article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
we would like to thank andy albrecht , richard matzner and ray rivers for extremely useful discussions .
l.m.a.b.s research was supported by j.n.i.c.t . under contract
. 9 t. w. b. kibble , j. phys . a 9 , ( 1976 ) 1387 .
b. zeldovich , mon . not .
192 , ( 1980 ) 663 ; a. vilenkin , phys .
lett . * 46 * , ( 1981 ) 1169 ; 46 , ( 1981 ) 1496(e ) .
h. b. nielsen and p. oleson , nucl .
b 61 , ( 1973 ) 45 .
e. p. s. shellard , nucl .
phys b 283 , ( 1987 ) 624 .
e. p. s. shellard , in : cosmic strings : the current status , eds . f. s. accetta and l. m. kraus ( world scientific , 1988 ) p. 25 .
r. a. matzner , comput .
2 , ( 1989 ) 51 .
laguna and r. a. matzner , phys .
letters 62 , ( 1989 ) 1948 .
l. m. a. bettencourt and r. j. rivers , imperial college preprint no .
imperial / tp/93 - 94/33 .
e. copeland and n. turok , unpublished .
p. j. ruback , nucl .
phys b 296 , ( 1988 ) 669 .
l. jacobs and c. rebbi , phys .
b 19 , ( 1979 ) 4486 .
e. meyers , c. rebbi and r. strilka , phys .
d 45 , ( 1992 ) 1355 .
p. a. shah , cambridge pre - print damtp 94 - 8 h. m. hodges and j. r. primack , phys .
d 43 , ( 1991 ) 3155 .
d. b. bennett and f. r. bouchet , phys .
60 , ( 1988 ) 257 ; in : the formation and evolution of cosmic strings , eds .
g. w. gibbons , s. w. hawking and t. vachaspati ( cambridge univ . press , cambridge , 1990 ) p. 359 .
t. vachaspati and a. vilenkin , phys .
67 , ( 1991 ) 1057 . | it is shown that for small relative angle and kinetic energy two type i @xmath0 strings can form bound states upon collision instead of the more familiar intercommuting configuration . the velocity below which this may happen
is estimated as function of the ratio of the coupling constants in the theory , crossing angle and initial kinetic energy .
cosmic strings formed at a grand unified symmetry breaking phase transition @xcite are a strong candidate to solve one of the most outstanding problems of cosmology namely that of structure seeding @xcite .
the standard cosmic string scenario assumes that strings after the phase transition rapidly approach the limit of zero width and become henceforth well described in terms of the nambu - goto action .
the other necessary ingredient to specify the evolution of the network is that at each collision strings should exchange ends or in other words should intercommute . in particular , whenever two segments of the same long string intersect a loop will be formed .
loops of string , in turn , are not globally topologically stable and will shrink by emitting their energy presumably as gravitational radiation .
this process allows for a faster decrease of the energy density in strings and precludes the possibility of a string dominated universe .
this whole construction rests crucially on the assumption that the outcome of a string collision always corresponds to intercommuting .
however , strings are in rigour classical non - trivial solutions of a variety of field theories displaying spontaneous symmetry breaking @xcite . the preferred model for realising string dynamics
, however , is the abelian higgs model @xmath1 in what follows we will also confine ourselves to it and its @xmath0 string solutions @xcite . the only way to establish intercommuting as the necessary outcome of the collision between two cosmic strings is to study the dynamics of the full field theory solutions .
this , however , involves solving a set of non - linear equations for the fields which proves to be a rather difficult task in most circumstances .
one way out is to perform simulations of collisions by solving the theory numerically .
this was done for several values of the ratio of the coupling constants in ( 1 ) , @xmath2 , supporting the conjecture that global , type ii and critically coupled strings ( i.e. , @xmath3 ) intercommute under essentially all circumstances @xcite .
similar studies regarding the behaviour of type i strings are much scarcer and seem to open the possibility for the existence of more complicated configurations @xcite . for a collision between two cosmic strings
to be analytically understood in terms of its underlying field theory one needs to look , in the first instance , at vortex dynamics .
indeed , both strings , at the collision point , can be projected onto a set of two orthogonal planes , thus reducing the problem to the study of vortex dynamics in two dimensions @xcite . in this picture
intercommuting results from the annihilation of a vortex - antivortex pair in one of these planes and the scattering at right angles of the vortex - vortex pair on the other .
the scattering at right angles , in turn , is only fully understood analytically in the critically coupled regime @xcite .
this important result may , although possibly under more severe restrictions , generalise to non - critically coupled vortices . as a practical result the interactions should vanish at zero separation as
is observed numerically @xcite , at least for @xmath4 not too far from unity .
this conjecture also seems to be well supported by several other numerical and semi - analytical recent studies of non - critically coupled vortex scattering dynamics @xcite .
it then seems likely that the effect of the interactions between cosmic strings should be felt at intermediate distances , i.e. , of the order of several scalar field coherence lengths .
the character of these is better known than their short distance counterpart . in particular , for @xmath5 ( type i ) @xcite these interactions are always attractive , regardless of the relative orientation of a pair of strings . in principle
this leads to the possibility of formation of a bound state ( characterised by the sum of the individual winding numbers of the two original strings ) as the outcome of a sufficiently favourable collision between two string segments . in this letter
we explore this scenario by investigating what configurations could compete with the usual intercommuting as the outcome of the collision of two straight type i strings at an angle and under which circumstances these can be expected to prevail .
a typical intercommuting event that occurs at the origin at @xmath6 may be taken to be between two strings whose initial configuration is given ( for @xmath7 ) by @xmath8 where @xmath9 .
note that @xmath10 is the velocity of approach .
we can write @xmath11 $ ] , where @xmath12 and @xmath13 . then , @xmath14 and @xmath15 if we have the normal type of intercommuting event , the solution for @xmath16 is easy to describe . for @xmath17 ,
the previous solutions are still valid , but in the region where @xmath18 and @xmath19 the partners have swapped over .
the solutions are as illustrated in figure 1 .
the `` horizontal '' pieces of string , in fig . 1 , are described by @xmath20 = ( t,\xi \sigma,\eta t , - \zeta \sigma),\ ] ] and @xmath21 = ( t,\xi \sigma,-\eta t , \zeta \sigma).\ ] ] note that the velocity on these sections is in the @xmath22-direction , not in the @xmath23-direction .
can we then find an alternative solution , based on the assumption that a segment of @xmath24 string is created at the moment of intersection ?
from symmetry it is easy to see that if such a segment is created it should be static and directed along the @xmath25-axis .
let us write @xmath26 , where @xmath27 is the string tension for unit winding number , and @xmath28 the corresponding value for @xmath24 .
@xmath29 is naturally a function of @xmath4 and , in particular for type i strings , we have @xmath30 . then the solution for the @xmath24 segment should be of the form @xmath31 it is reasonable to assume that the ends of this segment must move with definite velocity
let us take @xmath32 to be the solution for @xmath33 in the range @xmath34 , for some @xmath35 .
then the ends will move with speed @xmath36 ( obviously @xmath37 ) .
the overall shape must then be as illustrated in fig .
2 , with four links at the same places as before receding with speed @xmath38 , joined to the doubled section by new straight segments . we can then write down the expressions for the string position on each of the straight segments , since we know the coordinates of the end - points @xmath39 equivalently we can write , @xmath40 to satisfy the consistency conditions , we require @xmath41 ^ 2 + ( 1+\kappa)^2 \eta ^2 + ( 1+\kappa)^2 \zeta^2 = ( 1-\kappa)^2,\ ] ] or equivalently @xmath42 which yields @xmath43 thus , the value of @xmath35 is given by @xmath44 clearly a solution exists , if and only if , @xmath45 this is equivalent to requiring the angle between the strings , @xmath46 , to be small , or the velocity of approach , @xmath10 , to be small or both .
we have @xmath47 and consequently , ( see fig .
1 ) , @xmath48 the limit on the angle is then @xmath49 .
it is not surprising that a `` zipper '' can exist only for small angles and/or velocities . in particular , for @xmath50 , @xmath51 @xcite , and the constraint ( 15 ) becomes quite severe .
having shown that the formation of `` zippers '' is possible in the nambu - goto approximation we investigate how the consideration of the interactions between the two colliding strings alters the corresponding constraints on the collision parameters ( 15 ) . in the nambu - goto approximation we have seen that the difference between the tension of the @xmath24 string when compared to that of the two @xmath52 strings added together , allowed for the formation of the `` zipper '' . in this approximation
such a difference is restricted to the @xmath24 bridge only , disregarding the interactions between the two @xmath52 `` legs '' at each end of it . by including these interactions we
should then be able not only to render our former construction more realistic but also to relax the rather tight constraints it imposed . down to scalar coherence length separations
it was shown @xcite that the interaction energy between two @xmath52 string elements at an angle @xmath46 , is given approximately by @xmath53 , \ ] ] where @xmath54 and @xmath55 .
on the other hand each vortex possesses a self - energy @xmath27 , which naturally emerges as the string tension in the nambu - goto approximation and can be seen as its inertial mass for the study of collisions . for general couplings
@xmath27 has to be computed numerically . a best fit to its dependence on @xmath4 yields @xcite @xmath56 which should be valid to about @xmath57 accuracy , in the range @xmath58 we can then investigate the existence of bound states by matching the kinetic energy of two segments emerging from the collision , in the center - of - mass frame , to the potential energy of the system at a given separation . in so doing
we are assuming that only a finite length of string contributes to the dynamics , at the collision point .
we express this in terms of a simple cutoff , @xmath59 , a length along the string measured from the point of collision and take the velocity of each string as seen from the center - of - mass frame , @xmath60 , to be constant within such an interval , should be largest and then decreasing to zero smoothly with increasing distance along the string .
our procedure of introducing a simple cutoff @xmath59 , gives us , therefore , a measure of the total kinetic energy in the direction of separation , but not of its distribution along the string . ] .
this procedure yields @xmath61,\ ] ] which , in turn , leads to @xmath62 where @xmath63,\ ] ] and @xmath64 we see , in particular , that if we take @xmath59 to infinity the integral in ( 20 ) should rapidly converge to a constant while the inverse proportionality of the velocity relative to @xmath59 ensures that the system will always be free .
conversely , in the limit of vanishing @xmath59 , the picture for the collision of two vortices emerges , with the consequent drop in dependence on @xmath65 . for @xmath66 the integrand in ( 20 ) becomes a constant and this results trivially .
physically , @xmath59 should depend on the parameters of the collision , namely on the angle of approach @xmath46 and on the time - scale for the exchange of ends between the two strings
. determining @xmath59 would then require the detailed knowledge of the dynamics of the collision , which is unknown . in what follows we will , therefore , leave it as a free parameter and study the effect of its variation upon @xmath60 .
the distance @xmath67 corresponds to the separation at which the kinetic energy should match the potential energy exactly . because we expect expression ( 16 ) to break down at distances smaller than the coherence length of the scalar field
@xcite we take it to be @xmath68 .
this condition allows the system to scatter before the two segments are brought back together , under the effect of the interactions .
it should also correspond to a value close to the maximum of the interaction energy , since , for smaller separations , it was shown numerically @xcite that it should remain approximately constant as the separation vanishes , at least for @xmath4 not too far from unity .
figure 3 shows a plot of the dependence of the binding velocity , @xmath60 , on @xmath4 for four values of @xmath46 and @xmath69 .
we see the strong dependence on the angle of approach @xmath46 . in particular for orthogonal strings we expect the contribution from the potential energy to almost vanish and the system will be free . for intermediate angles ,
however , the binding velocity , @xmath60 can be naturally in the range of @xmath70 .
the actual value depends quite sensitively on @xmath59 .
figure 4 a ) , b ) , c ) show the dependence of the binding velocity on @xmath4 and @xmath59 for @xmath71 , respectively .
the dependence of the velocity on the angle of approach is shown in figure 5 , for @xmath72 and five values of @xmath73 .
in particular it becomes clear that the velocities only converge to a negligible value for @xmath46 close to @xmath74 . for intermediate angles
the variation is relatively small allowing for a large range of parameters for which the velocity is relativistic .
ultimately , one wants to relate the binding velocity ( 19 ) with the approach velocity of two typical strings in the network .
in particular the collision needs to be reasonably inelastic , since otherwise the system could be trivially time - reversed , resulting in configurations with the strings at the same separation and speed before and after the collision .
one way of parameterising this is by the use of an efficiency parameter , @xmath75 , in the following way . just before the two zeros of the scalar field superimpose , for very small separation
, we have argued that the system s energy should be essentially kinetic .
then the same follows for the configuration just after the collision .
however , the incoming energy will be channeled to all possible modes of an @xmath24 vortex , among which is the one corresponding to the separation of the two @xmath52 vortices .
we then take @xmath75 to be the ratio between the initial and final ( kinetic ) energies , in the separation mode , before and after the collision .
consequently , @xmath76 .
this , in turn , results in a relation between the initial and final velocities , before and after the collision @xmath77 where , @xmath78 will be taken to be the above @xmath60 .
once thus computed , @xmath79 can be related to a velocity at a given separation by taking the effect of the interactions into account , for a given spatial configuration .
the typical velocity quoted for a network of cosmic strings @xcite is about @xmath80 .
our estimates depend quite sensitively on two undetermined parameters @xmath59 and @xmath75 .
their values can , in principle , be found by looking at simulations or by solving the short distance dynamics of strings more exactly analytically .
if @xmath59 and @xmath75 then turn out to be large , only a very small fraction of all collisions will have a `` zipper '' as the outcome .
subsequent collisions of these configurations with other strings may lead to the peeling or `` unzipping '' of these higher winding number bridges as was shown in simulations by laguna and matzner @xcite .
as such the difference between the evolution of such a network and those evolving under the effect of intercommuting alone , should be negligible .
for sufficiently low values of @xmath59 and @xmath75 , however , we have shown that a large fraction of all collisions of strings in the network would lead to local higher winding number bridges that could grow as `` zippers '' .
the overall evolution of such a network can then be considerably different . in principle loop formation under these circumstances could still occur , at the same rate , but such loops would remain connected to their mother string .
this should have an effect on their subsequent evolution . a network of type i strings
can then be richer in string configurations than one which evolves simply by intercommuting at every collision .
this should have consequences for the evolution of the characteristic lengths in a string network , which , in turn , could reflect in the way strings seed energy distribution anisotropies , hopefully with potentially observable consequences both in structure in the universe and anisotropies in the cosmic microwave background radiation . |
diverse galaxy morphology is one of the cosmological mysteries in which the secret of structure formation and evolution might be hidden .
galaxy morphology is known to be correlated with luminosity and spatial environment , and there is a hope that knowledge on the correlations among them may provide clues to unravel the mystery .
this line of study began in the 1930 s .
hubble & humason ( 1931 ) found that clusters were dominated by ellipticals and lenticulars and that environmental factors played an important role in determining the morphology of galaxies .
oemler ( 1974 ) found the morphology - radius relation ; the late type galaxy fraction decreases with radius within a cluster .
this relation was confirmed by dressler ( 1980 ) who argued that the fraction of morphological types is a function of local galaxy density .
postman & geller ( 1984 ) extended this morphology - density relation down to the group environment . on the other hand ,
galaxy luminosity is also known to depend on environment .
an excess of bright galaxies in clusters has been pointed out by schechter ( 1976 ) .
it is now well - accepted that both amplitude and shape of the luminosity function of the individual morphological types vary with environment ( park et al .
2007 ) . despite these empirically known relations ,
it is very complicated to disentangle independent factors from the interrelations among various physical parameters of galaxies and environment . on the theoretical side ,
there have been many ideas about the mechanisms causing such observational effects .
gunn & gott ( 1972 ) argued that s0 galaxies in great relaxed clusters like coma were the result of spirals being stripped of gas by ram pressure stripping due to hot intra - cluster gas .
high speed encounters of galaxies with other halos typically in clusters cause impulsive heatings , called harassment ( moore et al .
1996 ) , and can transform spirals to early types .
strangulation ( balogh , navarro & morris 2000 ) is another mechanism that can also transform morphology through decline of the star formation rate ( sfr ) due to shut - off of the newly accreted gas when a galaxy enters a cluster or group environment and looses its hot gas reservoir .
there is also observational and theoretical evidence that the general tidal force field in clusters can transform infalling spirals to early types ( moss & whittle 2000 ; gnedin 2003 ) .
it is also plausible for galaxy morphology to be largely determined by the initial conditions when galaxies were formed .
gott & thuan ( 1976 ) argued that elliptical galaxies were produced by a larger initial density which would have higher density at turn - around and where star formation would be completed before collapse .
such larger initial density would be more likely in a region that would later turn into a high density environment .
it is important to note that all the nurture processes proposed so far are basically effective in group or cluster environments .
however , there seems to be morphology and luminosity - determining processes which work at all local densities in a continuous way .
this is supported by the fact that there is no feature in the galaxy property versus local density relation at any local density when morphology is fixed ( park et al .
but it was found that the fraction of late morphological type decreases sharply above the critical luminosity of about @xmath3 in the morphology versus luminosity relation ( choi et al .
2007 , hereafter paper i ) .
recently , the two degree field galaxy redshift survey ( 2dfgrs ; colless et al . 2001 ) and the sloan digital sky survey ( sdss ; york et al .
2000 ) have been extensively used to accurately measure the environmental effects on various physical properties of galaxies ( goto et al .
2003 ; balogh et al .
2004a , b ; blanton et al . 2005a ; croton et al .
2005 ; tanaka et al .
2004 ; weinmann et al .
2006 ; among many others ) . in this paper
, we extend our work in companion papers which studied the relations among various physical properties of galaxies in the sdss spectroscopic sample ( paper i ) , and the environmental dependence of the physical parameters ( park et al .
2007 , hereafter paper ii ) .
in particular , we will further inspect the findings that galaxy morphology varies with both the density measured on a few mpc scales and the distance to the nearest bright galaxy , and that the dependence of the early - type fraction on the nearest neighbor distance is strongest when the distance is about @xmath4 kpc .
it will be shown that this separation corresponds to the virial radius of the typical galaxies used in the study .
we will perform an extensive study of the dependence of galaxy morphology on luminosity , large - scale density , the local density due to the closest neighbor , and morphology of the neighbor .
galaxy morphology is found to depend on the various local environment parameters in a more complicated way .
taking into account the fact that galaxy morphology is closely related with luminosity , we extended our work to inspect the morphology transformation in conjunction with luminosity class transformation of galaxies .
for this purpose , we use large volume - limited samples divided into accurate morphological subsets of early and late type galaxies in several luminosity bins .
we use a large - scale structure sample , dr4plus ( lss - dr4plus ) , of the sdss ( york et al . 2000 ; blanton et al . 2003a ; fukugita et al . 1996 ; gunn et al .
1998 , 2006 ; hogg et al . 2001 ; ivezic et al . 2004
; lupton et al .
2001 ; pier et al .
2003 ; smith et al . 2002
; stoughton et al .
2002 ; tucker et al .
2006 ) from the new york university value - added galaxy catalog ( nyu - vagc ; blanton et al .
this sample is a subset of the sdss data release 5 ( adelman - mccarthy et al .
our major sample of galaxies used here is the lss - dr4plus sample referred to as `` void0 , '' which includes the main galaxies ( strauss et al .
2002 ) with the @xmath5-band apparent magnitudes in the range @xmath6 and redshift in the range @xmath7 . for the approximately 6% of targeted galaxies that lack a measured redshift because of fiber collisions ,
we assign the redshift of the nearest neighbor .
the @xmath5-band absolute magnitude @xmath8 used in this study is the ab magnitude converted from sdss magnitudes . to compute colors , we use extinction ( schlegel et al .
1998 ) and k - corrected model magnitudes .
the superscript 0.1 means the rest - frame magnitude k - corrected to redshift of 0.1 ( blanton et al .
all of our magnitudes and colors follow this convention , and the superscript will be subsequently dropped .
we also drop the @xmath9 term in the absolute magnitude .
we use the luminosity evolution correction of @xmath10 ( tegmark et al .
we adopt a flat @xmath11cdm cosmology with @xmath12 and @xmath13 .
completeness of the sdss is poor for bright galaxies with @xmath14 because of spectroscopic selection criteria ( which exclude objects with very large flux within the fiber aperture ) and difficulties of automatically measuring photometric properties of very extended sources .
as described in detail in section 2.1 of paper i , we add the missing bright galaxies and thereby extend the magnitude range by using various existing redshift catalogs which we match to the sdss data . in total , 5195 bright galaxies are added to the void0 samples within our angular sample boundaries ( see fig . 1 of paper ii ) .
volume - limited samples derived from the resulting catalog have nearly constant comoving number density of galaxies in the radial direction for redshifts @xmath15 .
we treat our final sample as effectively having no bright limit at redshifts greater than @xmath16 .
more details about this sample can be found in paper i. to study the effects of environment on galaxy properties it is advantageous for the observational sample to have the lowest possible surface - to - volume ratio , so that boundary effects are minimized ( see park et al .
[ 2005 ] regarding similar effects in a topology analysis ) .
for this reason we trim the dr4plus sample as shown in figure 1 of paper ii .
the three stripes in the southern galactic cap observed by sdss are not used because of their narrow angular extent .
these cuts leave approximately @xmath17 deg@xmath18 in the survey region . within our sample boundaries
, we account for angular variation of the survey completeness by using the angular selection function defined in terms of spherical polygons ( hamilton & tegmark 2004 ) , which takes into account the incompleteness due to mechanical spectrograph constraints , bad spectra , or bright foreground stars .
the resulting useful area ( with nonzero selection function ) within the analysis regions is 1.362 sr . in our study
it is very important to use volume - limited samples of galaxies that maintains uniform sampling in space and luminosity .
the absolute magnitude and redshift limits of a set of our volume - limited samples are shown in figure 2 of paper i. the definitions for the samples are summarized in table 1 .
we mainly use the @xmath19 sample which includes 74,688 galaxies with @xmath20 .
the @xmath21 sample , used in figure 5 , includes 80,479 galaxies with @xmath22 .
lcccc d3 & @xmath23 & @xmath24 & @xmath25 & 49571 + d4 & @xmath26 & @xmath27 & @xmath28 & 74688 + d5 & @xmath29 & @xmath30 & @xmath31 & 80479 + an important feature of this work is the use of large , accurate morphology subsets .
we first classify morphological types of galaxies using the prescription of park & choi ( 2005 ) .
galaxies are divided into early ( ellipticals and lenticulars ) and late ( spirals and irregulars ) morphological types based on their locations in the @xmath32 color versus @xmath33 color gradient space and also in the @xmath34-band concentration index space .
our study is about the morphological transformation between the early and late types , and not within the subclasses . to measure the color gradient and the concentration index , we use the @xmath35- and @xmath34-band atlas images and employ basic photometric parameters measured by the princeton / nyu group .
the boundaries between the two types in the three - dimensional parameter space is determined in such a way that the classification best reproduces the visual morphology classification .
the resulting morphological classification has completeness and reliability reaching 90% , as claimed by park & choi ( 2005 ) .
when photometry is excellent and galaxy images are well - resolved , more information from surface brightness fluctuations can be added for morphology classification .
but this is certainly not the case near the faint limit ( @xmath36 ) of the sdss sample we use .
our automatic classification scheme does not perform well when an early type galaxy starts to overlap with other galaxy .
this is because the scheme excludes galaxies with very low concentration from the early type class and blended images often erroneously give low concentration .
since we are investigating the effects of close pairs on galaxy luminosity and morphology , this problem in the automatic classification has to be remedied . in the case of the volume - limited sample d4
we perform an additional visual check of the color images of galaxies to correct misclassifications by the automated scheme for about @xmath37 galaxies located at the local density @xmath38 ( see below for the definition of @xmath0 ) . in this procedure
we changed the types of the rare galaxies which are blue but elliptical galaxies , blended or merging ones , or dusty edge - on spirals .
some non - sense objects are removed from the samples , and those with wrong central positions are corrected . to find relations between intrinsic physical properties of galaxies and their environment we require a well - defined and robust measure of environment .
we seek to use environmental parameters that are defined directly from observational data , and characterize the full range of galaxy environments , from the most massive clusters to voids .
most previous studies have used the local galaxy number density as a measure of local environment . in the present study
we will consider three kinds of environment .
one is the mass density described by many neighboring galaxies over a few mpc scale .
this is called the large - scale background density .
another is the local mass density attributed to the closest neighbor galaxy .
the third is the morphology of the closest neighbor galaxy .
the background density at a given location of a galaxy is measured by @xmath39 using the luminosity @xmath40 of the closest twenty galaxies in a volume - limited sample . here
the mass associated with a galaxy is assumed to be proportional to the luminosity of the galaxy .
the mean mass density within a sample of the total volume @xmath41 is obtained by @xmath42 where the summation is over all galaxies in the sample .
the mean density of each volume - limited sample is nearly a constant when the magnitude limit of the sample is fainter than @xmath43 .
only the relative mass to light ratios ( @xmath44 ) for early and late types are needed .
we assume @xmath45 at the same @xmath5-band luminosity .
this is our choice of the connection between luminosity and morphology with the host halo mass .
it is based on the fact that the central velocity dispersion of early type galaxies brighter than about @xmath46 is about @xmath47 times that of late types ( fig .
4b of paper i ) .
it is also consistent with our finding that the rms velocity difference between a galaxy and its neighbor galaxies is about 1.5 larger for the early types ( see section 2.4 ) .
we find the mean mass density @xmath48 where @xmath49 is the mass of a late - type galaxy with @xmath50 .
we use the spline - kernel weight @xmath51 for the background density estimation as in paper ii .
we vary the size of the spline kernel , @xmath52 , to include a fixed number of galaxies within the kernel weighting .
this adaptive smoothing kernel is often used in smoothed particle hydrodynamics simulations .
we use the spline kernel because it is centrally weighted , unlike the tophat or cylindrical kernel , and has a finite tail , unlike the gaussian .
because this smoothing kernel is adaptive and centrally weighted , it has the important feature that it does not oversmooth dense regions , thereby blurring the ` walls ' into the ` voids ' .
maps of the observed redshift space distribution of galaxies support this picture ( see fig .
3 of paper ii ) .
the adaptive nature of the smoothing kernel makes the physical smoothing scale systematically vary with environment , from a small smoothing scale in clusters to a very large smoothing scale in voids .
since the high density regions collapse while the under dense regions expand as the universe evolves , the adaptive kernel allows more uniform smoothing scale in the ` initial ' conditions compared to the method adopting a fixed - scale at the present epoch . in our study the number of galaxies required within the smoothing volume is set to @xmath53 , which we find is close to the smallest number yielding good local density estimates .
when searching for the spline radius containing the required number of galaxies , the smoothing volume can hit the sample boundaries or masked regions due to bright stars or galaxies .
in addition to this trouble , the angular selection function varies across the sky even within the survey region .
we first check if the location where the background density is being measured is within the survey area , and then ignore the survey geometry and angular selection function to search for the nearest 20 galaxies in order to find the spline radius and thus the local density using equation ( 1 ) .
we then randomly throw 1000 test particles into the smoothing volume around the location , and calculate the sum of normalized weights of the test particles assigned by the spline kernel , the angular selection function , and the radial boundaries .
the final local density is given by the first estimate divided by the sum of weights .
we have estimated the accuracy of the @xmath2 measure of the background density .
we identified dark halos with mass more than about @xmath54 ( kim & park 2006 ) from a large @xmath11cdm n - body simulation using @xmath55 particles made by park et al .
( 2005 ) , and measured the dark halo number density @xmath2 in real and redshift spaces .
the error in the redshift space @xmath2 is found to be 12 and 31% at densities @xmath56 and 10 , respectively . in paper ii
, only the @xmath57 galaxies with @xmath58 were used to trace the number density field .
the background density @xmath2 here differs from that of paper ii in that it represents the mass density rather than the number density and that the density tracers include all galaxies brighter than the faint magnitude limit of a sample . as a result
, @xmath2 distinguishes high density regions from low density regions much better .
when the @xmath19 sample is analyzed , the galaxies tracing the mass density are all galaxies brighter than @xmath43 .
the median radius of the smoothing kernel is @xmath59 mpc and its 68% limits are @xmath60 and @xmath61 mpc .
the small - scale density experienced by a target galaxy located at @xmath62 is estimated by @xmath63 where @xmath64 is the projected separation of the nearest neighbor galaxy from the target galaxy . the density due to
the nearest neighbor used in our work does not represent the small - scale galaxy number density , but rather the local mass density given by the nearest neighbor itself .
the method to find the nearest neighbor is described in the next section .
besides the nearest neighbor in terms of separation , we also adopted the neighbor galaxy giving the maximum value of @xmath0 or the galaxy causing the maximum tidal energy deposit per unit mass ( binney & tremaine 1987 ) .
we use a formula for the tidal energy deposit per unit mass @xmath65 where @xmath66 is the mass of the neighbor , @xmath67 is the velocity difference , and the size ( the rms radius ) of the target galaxy @xmath68 is set to 10 times the petrosian radius .
this combines the formulae for distant and penetrating encounters .
when we used these criteria to choose the neighbor , our results were essentially the same .
this is because the closest neighbors still dominate the latter two cases .
the fraction of neighbors selected differently in these three nearest neighbor selection schemes is only about 15% .
we therefore present our results using only @xmath0 of the nearest neighbor .
about 6% of galaxies in our samples missed the spectroscopic fibers due to the minimum fiber separation of 55@xmath69 per tiling in the sdss .
when we estimated @xmath2 or @xmath0 , we included those galaxies with borrowed redshifts from the nearest neighbor on the sky . to estimate the errors in @xmath2 and @xmath0 due to this treatment we artificially increased the fraction of galaxies with borrowed redshifts in our sample and compared the results with the original density estimates .
we first selected the galaxies who have measured redshifts and have neighbors within 55@xmath69 .
the redshift of the nearest neighbor which has a measured redshift is assigned to these galaxies , neglecting their own redshifts so that a total of 6% of all galaxies in the sample d4 received new redshifts by this procedure .
we found the ratio of @xmath2 calculated from the degraded sample to the original density had @xmath70 , and @xmath71 when @xmath72 , and @xmath73 , respectively .
this tells that assigning the nearest neighbor s redshift makes @xmath2 slightly overestimated in high density regions , and the scatter is between @xmath74% . this error is less than that due to the redshift space distortion .
we also checked the fraction of the nearest neighbors that were selected differently in the degraded sample .
it is found that 15 and 7% of the nearest neighbors are selected differently for the pairs with separations less than 0.1 and 1 @xmath75mpc .
for a given target galaxy with absolute magnitude @xmath76 selected from a volume - limited sample its nearest neighbor is found in the following way .
the closest neighbor is the galaxy brighter than @xmath77 with the smallest projected separation across the line of sight from the target galaxy and with a radial velocity difference less than @xmath78 . in the case of d4 containing galaxies brighter than @xmath46 ,
we study only those target galaxies brighter than @xmath79 so that their neighbors are complete .
these criteria are empirically found from the sample .
when we adopt galaxies fainter more than 0.5 magnitude as neighbors , our conclusions remain the same but our statistics become worse as the number of target galaxies becomes smaller .
it also reflects the fact that the effects of the neighbor are more prominent when the neighbor is brighter than the target galaxy . to determine @xmath78 we searched for all neighbor galaxies with velocity difference less than 1000 km s@xmath80 with respect to each target galaxy and with magnitude not fainter more than 0.5 .
figure 1 shows the velocity difference between the target galaxies with @xmath81 and neighbor galaxies ( in the d3 sample ) brighter than @xmath82 as a function of projected separation .
and their neighbors brighter than @xmath83 as a function of the projected separation .
most close neighbors have velocity difference less than 600 and 400 km s@xmath80 when the target galaxy is an early ( e / s0 ) and late ( s / irr ) morphological type , respectively .
the bottom panel for the late type galaxy indicates that contamination due the interlopers are serious when @xmath84 km s@xmath80 . ]
only the galaxies with measured redshifts are used .
it was found that the rms velocity difference of the neighbors is nearly constant out to the projected separation of @xmath85 kpc , and is 255 and 169 km s@xmath80 for early and late type target galaxies , respectively .
we adopt @xmath86 and @xmath87 km s@xmath80 for the early and late type target galaxies .
these limits correspond to about 2.3 times the rms values .
we first choose galaxies in the d4 sample with absolute magnitudes from @xmath79 to @xmath88 .
they are basically @xmath57 galaxies ( see table 2 of paper i for the schechter luminosity function parameters of the sdss galaxies ) .
the large - scale background environment is divided into high ( @xmath89 ) , medium ( @xmath90 ) , and low ( @xmath91 ) cases .
we then divide the small - scale neighbor environment into two cases when the closest neighbor is of early or late morphological type .
then we look at the morphology of the target galaxy as a function of the neighbor density @xmath0 when its luminosity , background density , and neighbor morphology are all fixed .
we first show figure 2 for an intuitive understanding of the effects of the neighbor s morphology and distance on the morphology of the target galaxy . at a fixed luminosity bin
the probability of a galaxy to be an early type is a sensitive function of the projected separation of the neighbor .
it also depends critically on the neighbor s morphology at separations shorter than about @xmath92 kpc
. however , at larger separations the dependence of @xmath93 on the neighbor morphology disappears . .
the solid line with filled dots is the case when the nearest neighbors are early types , and the dashed line with open circles is for the late ( s / irr ) type neighbor cases . ] to obtain a deeper physical understanding we now use the mass density due to the neighbor relative to the mean density of the universe as the abscissa of figure 3 . on the right side of the figure
the upper three curves are the cases when the nearest neighbor is an early type and the lower two curves are for late type neighbor cases .
the numbers next to the curves are the ranges of @xmath94 .
it can be immediately noted that the galaxy morphology is still a very sensitive function of the environment determined by the nearest neighbor even if both luminosity and the large - scale density environment are fixed . according to the morphology`density ' relation ( cf .
5 of park et al . 2007 ) galaxies are mostly of late type at very low ( large - scale ) densities and of early type at very high densities .
however , figure 3 tells that the majority of galaxies are of early type even at low densities if they have close early type neighbors , and that the majority of galaxies are of late type even at high densities if they are very isolated from other galaxies of comparable or brighter luminosity .
figure 3 tells us more facts .
when @xmath95 , the probability @xmath93 for a galaxy to be an early type does not depend on the neighbor s morphology .
its dependence on @xmath2 is also weak ( but discernable ) .
it is only the neighbor density @xmath0 ( besides luminosity ) which significantly affects @xmath93 ; the probability monotonically increases as @xmath0 increases .
hydrodynamic or radiative effects of the neighbor are not likely to be responsible for this increase because there is no neighbor morphology dependence .
the effects acting in the environments like cluster or group can not be responsible for the increase of early type fraction because the dependence on the large - scale background is insignificant in this range of @xmath0 . a possible mechanism responsible for this
is the tidal effects by the neighbor because the net effects of tide depend only on kinematic quantities such as separation , mass , size , and relative velocity ( eq .
tidal effects can accelerate the consumption of the cold gas in galaxies and tend to transform late types into early types
. it will be interesting to quantitatively estimate the strength of the tidal effects generated by a neighbor located at about the virial radius .
consider a galaxy having mass @xmath96 within radius @xmath97 .
its gravitational binding energy is @xmath98 .
the tidal energy deposit by a neighbor galaxy with mass @xmath66 and relative velocity @xmath67 is given by equation ( 5 ) times @xmath96 .
when the separation between the galaxies is much larger than the rms radius of the galaxy , the tidal energy deposit relative to the binding energy is given by @xmath99 if we adopt @xmath100 kpc , and @xmath101 100 km / s , we obtain @xmath102 .
therefore , the tidal effects between the dark halos of equal mass galaxies is not negligible at the separation of about the virial radius ( see below ) .
but the tidal force of the neighboring galaxy probably do little directly to the stellar and gas components of the galaxy because of small @xmath68 and @xmath97 for these components .
it remains to examine if the rearrangement of the dark matter in the halo during the distant encounter can perturb the embedded late type galaxy significantly enough to accelerate the consumption of its cold gas .
when @xmath103 , however , the probability at a fixed luminosity depends on all parameters , @xmath104 , and neighbor s morphology , in a complicated but understandable way . when galaxies have a late type closest neighbor , their @xmath105 deviates from that of the galaxies with early type neighbors at @xmath106 , and starts to drop at @xmath107 .
it is very important to note that this transition happens at the virial density .
the virialized density is @xmath108 relative to the mean density in the model universe we adopted ( gott & rees 1975 )
. then equations ( 3 ) and ( 4 ) give the radius of the virialized region @xmath109 and @xmath110 kpc for the late and early type galaxies with @xmath50 , respectively .
the neighbor galaxy is holding both cold and hot gases within the virial radius , and they can affect the fate of the galaxy falling into the region . if the neighbor is a late type galaxy , it is possible for its cold gas to flow into a trapped early type galaxy and transform it to a late type .
if this picture is true , then one would expect to find galaxies which appear to indicate hydrodynamic and/or radiative interactions between close pairs of galaxies before or without merger .
figure 4 shows a few examples .
a @xmath111 kpc scale bar at the rest frame is given to each panel .
the late type galaxies in the panels from ( a ) to ( e ) seem to be transferring cold gas to their early type companions . in panel
( a ) a long bridge extending from one spiral arm of a late type galaxy ( @xmath112 ) is connected to an early type galaxy ( @xmath113 ) . despite its morphology
the early type galaxy is very blue ( @xmath114 ) , and has extremely strong emission lines indicating active star formation .
the shape of this system has a surprising resemblance with a simulation made by cullen et al .
they presented the observed spatial and kinematic distribution of the atomic ( hi ) and molecular ( co ) gases in the spiral - elliptical interacting pair , arp 104 .
they also made a series of n - body simulations of the encounter between a spiral and an elliptical galaxy to reproduce a system such as arp 104 .
figure 8 of their paper is a best fitting case .
however , we note that the mass distribution in their simulation result looks even closer to the system in panel ( a ) .
an important lesson from their simulation is that a fraction of mass in the disc of the late - type galaxy can be transferred to the early - type component , and the spiral galaxy in panel ( a ) is probably feeding its disk material to its elliptical neighbor , which can be responsible for its spectral characteristics .
the panel ( b ) might be showing an example of the early - to - late transformation .
the galaxy vcc1748 ( binggeli , sandage , & tammann 1985 ) with the recession velocity of 11308 km s@xmath80 is at the lower right from the center .
it seems to be originally a red elliptical galaxy , but is now feathered with two blue spiral - looking arms . along the direction of its major axis is a blue spiral galaxy vcc1752 having almost the same recession velocity of 11299 km s@xmath80 .
the separation of two galaxies is only 34.5 @xmath75 kpc , and the local density felt by vcc1748 due to vcc1752 is @xmath115 relative to the mean density , much higher than the virial density .
one arm of the spiral galaxy is clearly connected to vcc1748 suggesting that the two arms are formed by the cold material flowed from vcc1752 along the bridge .
vcc1748 has previously been classified as sba , and seems to be an example that an elliptical galaxy transforms into a spiral when it closely encounters a gas - rich spiral without or before a merger event .
the systems from panels ( c ) to ( e ) also indicate mass transfer from a late type galaxy to an early type .
in fact , the mass transfer between galaxies during interactions has been widely studied both observationally and by numerical simulations .
simulations indicate that the detailed interaction features and the total mass transferred depend critically on the interaction parameters ( toomre & toomre 1972 ; sotnikova 1990 ; mihos & hernquist 1994 ) .
in particular , walin & stuart ( 1992 ) pointed out that the transfer of gas between galaxies can occur after they pass the point of perigalacticon .
observational evidence for mass transfer between galaxies has long been suggested by many studies .
knapp , turner , & cunniffe ( 1985 ) examined the hi content of elliptical galaxies and suggested that the gas has an external origin .
counterrotating disks or bulges in spirals and ellipticals with dust lanes can be interpreted as results of gas flows from the outside ( bertola et al .
1998 ; prada et al .
many examples of ongoing mass transfer between interacting galaxies have been studied .
some of them are arp 105 ( duc et al .
1997 ) , arp 194 ( marziani et al . 2003 ) , ngc 1409/1410 ( kell 2004 ) , and arp 104 ( cullen et al . 2007 ) .
the top three curves of figure 3 indicate that when the nearest neighbor is an early type , @xmath93 keeps increasing as @xmath0 increases above the virial density . in the light of the previous interpretation
this would be because the early type neighbor has a negligible amount of cold gas and both its hot gas and tidal effects tend to make the target galaxy falling within its virial radius an early type .
it is well - known that early type galaxies have extended x - ray halos ( ellis & osullivan 2006 ) . the x - ray emission and the hydrodynamic effects of hot gas can only accelerate consumption of cold gas in the close neighbor . since the bifurcation of @xmath93 for the late and early type neighbors occurs nearly at the virial density of the neighbor , the hydrodynamic effects seem more responsible for the morphology transformation than the radiative effects which should be less sensitive to a particular scale
. this may not be true for galaxies much fainter than @xmath50 , and the ionizing radiation from a bright elliptical galaxy could affect the cold gas content of faint spirals beyond the virial radius .
this seems to happen at least to the satellites of bright early and late type galaxies ( ahn , park , & choi 2007 ) .
the early type galaxies in panel ( f ) of figure 4 support the argument that the hot gas of an early type galaxy can remove cold gas from its neighbor ( in this case we have redshift only for the middle one .
we assume the two early type galaxies form a real interacting system .
they have nearly the same photometric redshifts ) .
the galaxy in the middle has a blue arc on the side facing the spiral galaxy .
however , the galaxy does not have such an arc on the opposite side facing brighter elliptical galaxy .
we speculate that the hot gas of the elliptical neighbor at the bottom may be responsible for non - existence of the second arc of the middle galaxy as an elliptical neighbor can ionize or push away the cold gas of its neighbor galaxy by a direct physical contact and make it look like an early type ( see sofue 1994 for a demonstration of such process in simulations ) .
we conclude that the physical properties of close neighbors are of paramount importance in the fate of a galaxy .
one can also learn from figure 3 that @xmath93 is systematically higher for higher @xmath2 when @xmath0 is higher than the virial density .
the mechanisms suggested for morphology transformation to early type due to large - scale environment include the ram pressure stripping of a late type galaxy falling into a cluster holding hot gas ( gunn & gott 1972 ) , removal of hot gas reservoir and stopping of cold gas supply ( strangulation ; larson et al .
1980 ) , or impulsive heating due to numerous encounters with other galaxies in high density environment ( harassment ; moore et al . 1996 ) .
even though these mechanisms may be acting in the very high large - scale density regions , none of these can be fully responsible for the large - scale background dependence we found because they all predict that @xmath93 is enhanced as @xmath2 increases independently of @xmath0 , which contradicts figure 3 .
( note also that we can not resolve the virialized region in the large - scale density @xmath2 . ) the fact that the strong @xmath2 dependence is observed only when @xmath1 , suggests that the effect of the background is indirect , and the mechanism may be associated with a physical property of the closest neighbor which in turn depends on the large - scale environment .
for example , this dependence can arise if the density and temperature of the hot gas held by galaxies are higher in higher density environments .
such a trend has been found only for the brightest group or cluster elliptical galaxies .
they are reported to have higher x - ray luminosity at given optical luminosity in higher density regions ( osullivan et al .
2001 ) . since it is very important to know the dependence of the x - ray emitting hot gas of normal early type galaxies on the large - scale environment to understand the dependence of @xmath93 on @xmath2 , we investigated the relation between x - ray luminosity of early types and the large - scale background density .
we used x - ray data from the @xmath116 all - sky survey ( voges et al .
1999 ) positionally matched with galaxies in the sdss dr4plus catalog . among the matched sources there are 78 early types brighter than @xmath43 within the volume of d4 sample , we find that there exists a clear positive correlation between the x - ray luminosity and @xmath2 at fixed optical ( @xmath5-band ) luminosity .
our accurate density estimator enabled us to detect this correlation .
the dependence of @xmath93 on @xmath2 at @xmath1 can be explained by this correlation .
the dependence is indirect and becomes effective only when the hot gas of the neighbor galaxy starts to be influential at separations between galaxies less than the virial radius .
so far , our study is done for the case when the luminosity of the target galaxies is restricted to @xmath117 .
figure 5 shows how the morphology transformation relation scales with luminosity . from @xmath118 to @xmath88 , and having early ( filled circles ) and late ( open circles ) neighbors .
the upper four curves used d5 sample . ] the bottom two ( dotted ) curves are the same as those in figure 3 but summed over all large - scale background densities . the early - type fraction @xmath93 monotonically increases as @xmath0 increases when the neighbor is an early type , but drops at @xmath119 when the neighbor is a late type .
the remaining curves are obtained from a deeper and brighter volume - limited sample of galaxies with @xmath120 ( sample d5 in paper i ) .
the middle two curves are for galaxies with @xmath121 having an early ( higher @xmath93 at high @xmath0 ) or late - type closest neighbor .
the top two curves are for the half magnitude brighter galaxies with @xmath122 .
we first note that the bifurcation of the early and late type neighbor branches occurs at the same @xmath123 .
the neighbor s morphology is still an important environmental factor in the morphology transformation at these brighter magnitudes .
figure 5 tells that the whole morphology transformation relations as a function of @xmath0 and neighbor s morphology , scale monotonically with luminosity .
we have also confirmed this scaling at the fainter magnitude range from @xmath43 to @xmath118 . above @xmath124 a similar scaling
is seen but the scaling amplitude is suddenly very large .
this is because the fraction of early type galaxies increases steeply above the magnitude of @xmath125 ( see fig .
8 of paper i ) .
they are often the central galaxies in groups and clusters .
to summarize , a late type galaxy tends to become an early type due to the tide exerted by its closest neighbor when it is separated by more than the virial radius of the neighbor . at separations
less than the virial radius the galaxy still tends to transform to an early type if the neighbor is an early type .
this trend is accentuated at higher large - scale background density .
but if the neighbor is a late type , the direction of morphology transformation tends to reverse and an early type can become a late type as it approaches its bright spiral neighbor . when one considers galaxies of different luminosity , all behaviour remain the same and the probability for a galaxy to be an early type simply scales with luminosity over the absolute magnitude ranges we study .
therefore , our understanding of galaxy morphology can not be complete unless we understand how this luminosity morphology relation originates . from figure 3
we were able to propose a number of physical processes playing roles in the galaxy morphology transformation .
however , one thing that figure 3 can not tell us is what happens when galaxies ( located at very high @xmath0 ) finally merge . after a galaxy undergoes a major merger with its closest neighbor , the next closest one will become the new closest neighbor .
the luminosity or mass of the merger product will be higher than those of the merging galaxies , and the small - scale density @xmath0 will drop greatly as the next closest neighbor as bright as the merger product will be usually far away .
if the galaxies had cold gas , the star formation must become relatively higher during the merger and the total cold gas content in the merged one will be less than the sum of the cold gas before the merger due to merger - induced starbursts . the next major merger will occur with a more massive galaxy ( as massive as the merged product ) which tends to have a lower fractional mass of cold gas . as this approach - merge process repeats , the fractional mass of the merger product in the form of cold gas will eventually vanish .
this scenario can explain why almost all most massive galaxies are early types .
the scenario can be supported by our finding that galaxy luminosity depends on @xmath0 at a fixed large scale density .
when one inspects the effects of @xmath0 or the nearest neighbor distance on various physical parameters of galaxies , one finds only a very weak signal for all parameters except for luminosity ( y .- y .
choi et al .
2007 , in preparation ) . figure 6 shows the absolute magnitudes of galaxies in @xmath19 as a function of @xmath0 .
the upper panel is for early type galaxies , and the bottom panel is for late types . as in figure 3
, we divide the large - scale background mass densities of galaxies into three bins according to their @xmath2 .
we plot only those galaxies sitting at the highest @xmath2 ( red dots ) or at the lowest @xmath2 ( blue dots ) bins to avoid overcrowding .
three curves are the median magnitudes of galaxies in three @xmath2 bins as a function of @xmath0 , and the curves for early types are repeated in the bottom panel as dashed curves .
one can immediately notice that the luminosity of galaxies brighter than @xmath50 and located in high @xmath2 regions is much higher at smaller @xmath126 .
namely , galaxies are brighter as they are more isolated . and
it is a strong function of @xmath0 when @xmath127 .
but it can be also noticed that luminosity is nearly independent of both @xmath0 and @xmath2 when @xmath128 , and that all these behaviours are almost the same for both early and late type galaxies ( but early types are slightly brighter than the late types in the magnitude range we explore ) .
we find the same behaviour of luminosity when fainter galaxies ( @xmath20 ) are included .
to explain why isolated galaxies are brighter in high @xmath2 regions and become faint as @xmath0 increases , one can argue that brighter galaxies preferentially form in higher large - scale density regions and that galaxies become fainter by tidal stripping as they approach each other .
however , this scenario is contradictory with the fact that the median luminosity is almost the same in different @xmath2 regions at @xmath128 , and that galaxy luminosity is independent of @xmath0 at very high @xmath0 where the tidal stripping should be more active .
an important fact that should be noted in figure 6 is that galaxies become more isolated ( small @xmath0 ) when @xmath2 is smaller or when luminosity is higher . in high @xmath2 regions galaxies are on average much closer ( higher @xmath0 ) to one another than those in low @xmath2 regions .
inevitably , the merge rate is relatively higher in higher @xmath2 regions .
we have visually identified merging ( or close encounter ) galaxies in the d2 sample brighter than @xmath50 .
we looked for strongly interacting galaxies that show distorted bodies with connecting bridges and/or trails , and have magnitude difference less than 1.5 .
we included the cases that an early type galaxy is seen within the optical image of the other early type .
we found , in the three density bins of @xmath129 , @xmath130 , and @xmath131 , the fractions of galaxies undergoing major merger / interaction are 3.0 , 4.8 , and 6.8 % , respectively .
this confirms the strong dependence of the major merger / interaction rate on @xmath2 ( d. g. kim et al .
2007 , in preparation ) .
the galaxy merger rate in high density regions must be higher than our estimate because there are more early type galaxies at high densities and the merge between early types is more difficult to notice as merging early types are less likely to show bridges or trails than late types .
when a galaxy merges with another of equal mass , the merger product will have twice the mass if the mass loss during the merger is negligible . since galaxies with twice the mass of the merging galaxies will be rarer
, it will be more difficult for the merger product to find a neighbor that can significantly influence its morphology and luminosity .
therefore , the merger product will jump to the upper ( brighter ) left ( low @xmath0 ) part of each panel in figure 6 , as we argued before .
the rapid brightening of galaxies at @xmath132 in the case of the highest @xmath2 bin , may be due to this process . at what value of @xmath0
will the galaxy find itself after merger ?
it is reasonable to predict that the new @xmath0 is typically less than @xmath133 .
if new @xmath1 , it implies two merging galaxies were already within a virial radius of a third one of about twice mass before merger .
this is not very likely because the more massive one in general impedes their merger by tidal force and tends to swallow the little ones before they merge unless the pair were already ready to merge before it fell into the more massive one .
this can explain why luminosity is nearly constant for galaxies at @xmath1 .
recently - merged galaxies usually do not fall at such @xmath0 .
therefore , the merger hypothesis can explain all features of figure 5 . to find more dynamical evidence for this scenario we inspected the late type galaxies showing indications of recent merger events .
our prediction is that there are more galaxies showing post merger features at @xmath134 for a given luminosity and a fixed large scale density . for late type galaxies we define post merger features as large displacement of the galaxy nucleus from the center , turmoil features , and/or very close double cores .
the selection is subjective , but is made only for quite clear cases .
only late types are used for higher reliability . for galaxies with @xmath135 and located in the highest large - scale background density region ( @xmath136 ) the fractions of galaxies showing the post merger features are @xmath137 and @xmath138% when @xmath139 and @xmath140 , respectively .
for more luminous galaxies with @xmath141 the fractions increase to 10.8 and 4.0 % , respectively . therefore , there are more recently merged galaxies among isolated ones than among those with close neighbors .
figure 7 illustrates a merger in an isolated environment . the @xmath5-band magnitude and redshift
are given to all galaxies brighter than @xmath142 .
at the center of the picture is a spiral galaxy ngc3695 with two cores indicating that it is at the ending stage of merger .
ngc3695 has @xmath143 , and is a blue spiral galaxy showing very strong star formation activity .
nearly equal brightness of the two nuclei implies that two galaxies of comparable mass have merged .
since ngc3695 is completing merger , the next closest galaxy will become a new neighbor .
the bright galaxy at the lower left of the picture is ngc3700 .
because ngc3700 has velocity difference of 12,300 km s@xmath80 , it does not qualify as a neighbor .
the closest neighbor is found at the separation of @xmath144 mpc , and ngc3695 is a good example of very isolated galaxy which once was a close pair of galaxies . to summarize
, galaxies travel from the left to the right of figure 6 at a given luminosity . during this travel
, galaxies are affected by the morphology transformation processes suggested by figure 3 , and can jump from one panel to the other .
when they reach very high @xmath0 , they undergo a merger and become brighter .
as they become brighter through mergers , they will find themselves more isolated from their neighbors of comparable mass .
as galaxies repeat such travel , the fractional mass in cold gas will decrease and galaxies with higher mass will tend to be of early type .
a full understanding of the origin of galaxy morphology will be obtained if one finds the initial distribution of morphology and the rule of morphology transformation .
the present study addresses the second issue . if galaxies transform their morphology and luminosity in accordance with the unified scenario we propose , a number of galactic phenomena found in the past can be qualitatively explained .
because the merger rate is higher in the higher large - scale density regions , there are more bright merged galaxies in high density regions , which can explain the luminosity - density relation ( paper ii ) .
as the merger process repeats , galaxies become brighter but also tend to become early types because the cold gas fraction decreases .
this can explain the luminosity morphology relation .
furthermore , the effect of hot gas of the neighbor on a galaxy orbiting within the virial radius tends to make the galaxy an early type and its effects are stronger in high large - scale density regions .
this can explain the morphology density or morphology(clustercentric ) radius relation ( more discussion is given below ) .
in addition to these important relations our picture can explain the following . 1 .
_ holmberg effect _ holmberg ( 1958 ) discovered that the color of paired galaxies are closely correlated .
we predict that as two galaxies approach each other and enter into the virial radius , hydrodynamic and radiative interactions of cold and hot gases occurs .
the primary galaxy tends to make the pair system have common properties .
the color of the secondary can become closer to that of the primary .
2 . _ star formation rate density correlation _
the fraction of star - forming galaxies is known to depend strongly on the local density .
balogh et al . ( 2004a ) showed that the fraction of @xmath145 emission - line galaxies is lower in regions that are overdense at both 3.85 and @xmath146 mpc scales . to a large extent
this correlation is associated with the morphology density relation .
our picture predicts that the cold gas content and thus the sfr in galaxies will be lower in higher large - scale density regions where the merger rate is higher and the early - type fraction is higher . at the same time , the morphology of a galaxy is sensitive to that of its neighbor when it is located within the virial radius of the neighbor .
the sfr must be so too .
the neighbor s morphology depends on the background density when the pair separation bin is small than the virial radius . therefore , the sfr should appear dependent on both small and large - scale densities . when the large - scale density and neighbor morphology are fixed ,
the sfr of galaxies in close pairs is predicted to be depressed for the early type neighbor case or enhanced for late type neighbor case .
conformity in galaxy morphology type _
this is essentially the same as the previous two relations .
it has been observed both in group / cluster environment and at galaxy scales .
weinmann et al .
( 2006 ) reported that a late - type central galaxy has a significantly higher fraction of late type galaxies within its halo than an early - type central galaxy of the same mass ( see also hickson et al .
1984 and ramella et al . 1987 )
. this can be just because of the strong hydrodynamic and radiative influences of the bright central galaxy on the less bright surrounding galaxies which are orbiting within the virial radius of the central one .
this galaxy morphology conformity is also observed within the halos of individual galaxies .
an elliptical galaxy is found to host a swarm of early type satellite galaxies while a gas - rich spiral galaxy tends to hold many late type satellites ( h. ahn et al .
2007 in preparation ) .
this is a scaled - down version of the phenomena observed in groups and clusters , and can be consistently explained by our picture .
morphology density versus morphology radius relation _
galaxy morphology depends strongly on the clustercentric radius ( melnick & sargent 1977 ) or local density of galaxies in clusters ( dressler 1980 ) .
it would seems impossible to find which of the factors is the more fundamental .
our picture claims that it is the local density which really controls the phenomena .
the large - scale background density determines the average distance to the nearest neighbor and the merger rate which in turn determines the mean brightness and morphology of the neighbor .
the small - scale density due to the nearest neighbor determines the strength of the tidal and hydrodynamic effects .
the virial radii of the group / cluster a galaxy belongs to , or of its neighbor galaxy are statistically determined by local densities .
when a galaxy enters within the virial radius of the group / cluster or of the neighbor galaxy , it starts to interact with the hot / cold gas trapped within the radius .
therefore , the virial radius has a fundamental importance in the evolution of galaxy morphology .
_ morphology - dependent large - scale clustering _
it is well - known that the spatial distribution of the early - type galaxies has higher clustering amplitude than the late types .
this can be because the early types preferentially form in high density regions and their clustering is more biased than the late types relative to the mass .
however , our results imply that even if the initial morphology of the newly - born galaxies is of late type , the luminous early type galaxies are formed preferentially in high density regions because the binary interaction is more frequent , the merger rate is higher , and the cold gas consumption is faster there .
once the early type galaxies with hot gas appear in high density regions , they will promote the transformation of morphology of neighboring spirals through galaxy - galaxy interaction .
since high density regions are compact , this will result in the morphology - dependent clustering amplitude .
our picture on the transformation of galaxy morphology and luminosity also explains some recent findings and can make predictions on galaxy properties . 1 .
_ existence of early types in underdense regions _ in our picture
early type galaxies can form in low density regions as well as in high density regions but at a lower efficiency through infrequent pair interactions and mergers .
it can be inferred from figure 6 which shows there are much fewer galaxies with @xmath147 in low density regions ( blue points ) .
it predicts that , in underdense regions where there are more gas - rich late - type neighbors , the very bright early types will be very rare and the fraction of blue star - forming early types will be higher .
this is exactly what is seen in figure 11 and 12 of paper ii .
there are ` normal ' ellipticals even in very underdense regions as demonstrated by the fact that the red - sequence of early types hardly changes as the large - scale density changes .
however , in underdense regions the red - sequence has a much shorter extension toward the bright magnitudes and the fraction of blue galaxies increases .
the fractions of early - type galaxies brighter than @xmath148 among those with @xmath149 are 9.4 , 6.6 , and 2.8 % in the regions where @xmath150 , @xmath90 , and @xmath151 , respectively .
on the other hand , the fractions of blue early type galaxies with @xmath152 and @xmath20 are 1.2 , 2.3 , and 3.4 % in those three regions .
properties of early type galaxies in low density regions are also studied by croton et al .
( 2005 ) and rojas et al .
the simple scenario that postulates formation of all early type galaxies in initially high density regions has a difficulty in explaining existence of the normal and blue ellipticals in very low density regions .
2 . _ isolated galaxies _
isolated galaxies are often studied in the hope that the evolution of galaxies can be understood with no contamination from environment - related processes .
isolated galaxies are usually defined as those which do not have companions brighter than a certain magnitude difference within a certain projected distance and radial velocity difference .
they are considered to be passively evolving galaxies formed in isolation via gravitational collapse of a primordial protogalactic cloud ( marcum , aars , & fanelli 2004 ) .
however , our scenario drawn from figure 6 predicts the most isolated galaxies are likely to be those that experienced a major merger relatively recently and whose internal evolution was greatly disturbed .
this is particularly so in high density regions or for very bright galaxies .
the wish for freedom from environmental effects will be approximately fulfilled only for the galaxies isolated at very low densities , namely for the void galaxies .
for example , marcum et al .
( 2004 ) have selected extremely isolated elliptical galaxies from the catalog of isolated galaxies ( karachentseva 1973 ) and found that four out of nine showed merger signatures .
it should also be noted that isolation of a galaxy is strongly correlated with the large - scale density and characteristics of isolated systems can be easily confused with those of underdense environment ( i.e. varela et al . 2004 ) .
redshift evolution of morphology and luminosity _
our scenario is consistent with the common wisdom that galaxies were on average less massive in the past and that the late - type fraction of galaxy morphology was higher at higher redshifts .
it predicts that galaxies located in high density regions should show relatively large differences in luminosity , color , sfr , and morphology as redshift changes because the mean separation between galaxies is smaller and the merger rate is higher there .
the fraction of early type galaxies should monotonically decrease and the fraction of blue early type among early types should increase at higher redshifts . however , the energetic radiation from numerous active galaxies , which are a minor population in our sample , might have critical roles in the evolution at high redshifts .
much less evolution will be observed for galaxies in low density regions where the mean galaxy separation is large because the interaction between galaxies is weaker and merger is less likely .
correspondingly , the star formation density relation will change in such a way that the strongest sfr of galaxies be observed in higher density regions as redshift increases ( see elbaz et al .
2007 for an observational detection of this phenomenon ) .
our scenario is also consistent with the recent finding from a n - body simulation that internal physical properties and spatial clustering of dark halos depend not only on the mass but also on the assembly history ( gao , springel , & white 2005 ; wechsler et al .
2006 ; gao & white 2007 ; croton , gao , & white 2007 ) . at high redshifts ,
galaxies at high densities are expected to be less depleted in cold gas and experience more frequent interactions than those at low densities . since at higher redshifts the space was compressed but there were fewer number of massive galaxies , the merger rate of objects above a given mass
is expected to evolve rather slowly ( see lotz et al . 2006 and bell et al .
2006 for observational evidence ) the galaxy morphology can also evolve due to the infalling gas in the form of intergalactic medium or dwarf galaxies . in less massive halo ,
the disk can be formed by cold flows ( @xmath153k ) , and in more massive halos the infalling gas is first heated by shocks to near the virial temperature ( @xmath154k ) before it is accreted to form a disk ( dekel & birnboim 2006 ; keres et al .
the relative importance between the galaxy - galaxy interaction and the gas / dwarf accretion on morphology transformation will depend on the large - scale environment and the time scales of gas cooling and interaction between galaxies .
for examples , the importance of the infalling gas will be relatively higher in low density regions where the interaction rate is lower and there are more gas rich faint objects .
it is also possible for galaxies to transform their morphology between early and late types through secular internal processes like exhaustion of cold gas and aging of the stellar population .
the time scale of such secular evolution is probably longer than those of galaxy interaction or gas accretion before the present epoch
. it will be interesting to compare the relative importance of all these processes in different environments .
we have analyzed a set of volume - limited samples drawn from the sdss data in order to understand the morphology luminosity environment relation .
our previous study on the environmental dependence of galaxy properties ( paper ii ) showed that most of the environmental dependence of physical parameters disappears once luminosity and morphology are fixed and that the morphological type of galaxies is affected by both small - scale ( due to the nearest neighbor ) and large - scale densities .
in particular , a strong dependence of morphology on the nearest neighbor distance was discovered when the distance was about @xmath4 kpc . for galaxies with @xmath82 , which is the faint limit of the sample d2 used in paper ii
, the critical distance of @xmath4kpc corresponds to the neighbor density of @xmath155 or 1040 for late or early type neighbor , respectively .
this essentially corresponds to the virial density . in the present work , based on the lessons from the previous study , we divided the environment into the large - scale background density , small - scale density due to the nearest neighbor , and the nearest neighbor s morphology .
our major findings are the following .
1 . transformation of galaxy luminosity and morphology classes is going on in all large - scale density environments through interactions and mergers between close galaxy pairs .
2 . at galaxy pair separations larger than the virial radius of the neighbor galaxy ,
the morphological type of a galaxy is nearly independent of the neighbor s morphology .
it is argued that the tidal effects of the neighbor transform late type galaxies into early types .
it is also hardly affected by the general tidal force field as indicated by its insensitivity to the large - scale background density .
3 . at separations from the closest neighbor shorter than the virial radius , the probability of a galaxy to be an early morphological type continues to rise as @xmath0 increases if the neighbor is an early type . the tidal , hydrodynamic , and radiative effects can result in this phenomenon .
if the neighbor is a late type , however , the probability decreases as @xmath0 increases and the galaxy tends to become a late type .
the tide and the hot gas may be competing with the cold gas flowing from the neighbor in this circumstance .
the early type probability as a function of neighbor morphology , large- and small - scale densities , scales systematically as the luminosity changes over the magnitude range from @xmath156 to @xmath157 .
the fraction increases sharply for @xmath158 .
it is found that galaxy luminosity depends sensitively on the local mass density due to the nearest neighbor .
isolated galaxies tend to be brighter and are more likely to be recent merger products .
the merger - driven transformation of luminosity class can result in these phenomena .
we present a unified scenario that the galaxy morphology and luminosity classes change as galaxies approach one another and undergo a series of ( mainly two - body ) interactions and mergers . as the average interaction and merger rates
are a function of the background density , the morphology luminosity density relation naturally arises under this scenario .
we were not able to study what happens to galaxy morphology and luminosity when the large - scale background density @xmath2 exceeds the virial density .
this is because the smoothing scale used to calculate @xmath2 , which is @xmath159 mpc , is larger than the size of the virial radius of clusters , which is typically about @xmath144 mpc . in a cluster environment
, we expect to see morphology transform of late type galaxies when they are within the virial radius of the cluster .
we emphasize that one needs to take into account factors other than just one local density estimate when the effects of galaxy interactions in different environments are studied .
in particular , the morphology of the closest neighbor is important when the separation is smaller than the virial radius of the neighbor . without distinguishing the nearest neighbor by morphological type ,
the study of the effects of interactions on color , sfr , or morphology fraction will find almost a null signal because the early and late neighbors will tend to induce opposite trends .
we will further investigate such effects of galaxy interactions in a forthcoming paper ( y .- y .
choi et al .
2007 in preparation ) .
the authors thank h. b. ahn , j. p. ostriker , and m. s. vogeley for helpful comments .
cbp acknowledges the support of the korea science and engineering foundation ( kosef ) through the astrophysical research center for the structure and evolution of the cosmos ( arcsec ) .
jrg is supported by nsf grant ast 04 - 06713 .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
, adelman - mccarthy , j. k. , et al .
2007 , , submitted , ahn , h. , park , c. , & choi , y .- y .
2007 , , submitted , balogh , m. l. , et al .
2004a , , 348 , 1355 , balogh , m. l. , baldry , i. k. , nichol , r. , miller , c. , bower , r. , & glazebrook , k. 2004b , , 615 , l101 , balogh , m. l. , navarro , j. f. , & morris , s. l. 2000 , , 540 , 113 , bell , e. f. , phleps , s. , somerville , r. s. , wolf , c. , borch , a. , & meisenheimer , k. 2006 , , 652 , 270 , bertola , f. , corsini , e. m. , vega beltran , j. c. , pizzella , a. , sarzi , m. , cappellari , m. , & funes , j. g. 1999 , , 519 , 127 , binggeli , b. , sandage , a. , & tammann , g. a. 1985 , , 90 , 1681 , binney , j. , & tremaine , s. 1987 , galactic dynamics ( princeton : princeton university press ) , 438 , blanton , m. r. , eisenstein , d. , hogg , d. w. , schlegel , d. j. , & brinkmann , j. 2005a , , 629 , 143 , blanton , m. r. , et al .
2005b , , 129 , 2562 , blanton , m. r. , lin , h. , lupton , r. h. , maley , f. m. , young , n. , zehavi , i. , & loveday , j. 2003a , , 125 , 2276 , blanton , m. r. , et al .
2003b , , 125 , 2348 , choi , y .- y . ,
park , c. , & vogeley , m. s. 2007 , , 658 , 884 ( paper i ) , colless , m. , et al .
2001 , , 328 , 1039 , croton , d. j. , et al .
2005 , , 356 , 1155 , croton , d. j. , gao , l. , & white , s. d. m. 2007 , , 374 , 1303 .
cullen , h. , alexander , p. , green , d. a. , & sheth , k. 2007 , , 376 , 98 , dekel , a. , & birnboim , y. 2006 , , 368 , 2 , dressler , a. 1980 , , 236 , 351 , duc , p .- a . , brinks , e. , wink , j. e. , & mirabel , i. f. 1997 , , 326 , 537 , elbaz , d. , et al . 2007 , , 468 , 33 , ellis , s. c. , & osullivan , e. 2006 , , 367 , 627 , fukugita , m. , ichikawa , t. , gunn , j. e. , doi , m. , shimasaku , k. , & schneider , d. p. 1996
, , 111 , 1748 , gao , l. , springel , v. , & white , s. d. m. 2005 , , 363 , 66 , gao , l. , & white , s. d. m. 2007 , , 377 , l5 , gnedin , o. y. 2003 , , 589 , 752 , goto , t. , et al .
2003 , , 346 , 601 , gott , j. r. , & thuan , t. x. 1976 , , 204 , 649 , gott , j. r. , & rees , m. j. 1975 , , 45 , 365 , gunn , j. e. , et al .
2006 , , 131 , 2332 , gunn , j. e. , et al .
1998 , , 116 , 3040 , gunn , j. e. , & gott , j. r. 1972 , , 176 , 1 , hamilton , a. j. s. & tegmark , m. 2004 , , 349 , 115 , hickson , p. , ninkov , z. , huchra , j. p. , & mamon , g. a. 1984 , 111 , 367 , hogg , d. w. , finkbeiner , d. p. , schlegel , d. j. , & gunn , j. e. 2001 , , 122 , 2129 , holmberg , e. 1958 , lund medd .
obs . ser .
ii , 136 , 1 , hubble , e. , & humason , m. l. , 1931 , , 74 , 43 , ivezi , z. , et al . 2004 , an , 325 , 583 , keres , d. , katz , n. , weinberg , d. h. , & dave , r. 2005 , , 368 , 2 , keel , w. c. 2004 , , 127 , 1325 , kim , j. , & park , c. 2006 , , 639 , 600 , knapp , g. r. , turner , e. l. , & cunniffe , p. e. 1985 , , 90 , 454 , karachentseva , v. e. 1973 , astrof .
issledovanija byu . spec .
8 , 3 , larson , r. b. , tinsley , b. m. , & caldwell , c. n. 1980 , , 237 , 692 , lotz , j. , m. , et al .
2006 , submitted ( astro - ph/0602088 ) , lupton , r. h. , gunn , j. e. , ivezc , z. , knapp , g. r. , kent , s. , & yasuda , n. 2001 , in asp conf .
238 , astronomical data analysis software and systems x , ed .
f. r. harnden , jr .
, f. a. primini , & h. e. payne ( san francisco : asp ) , 269 , marziani , p. , dultzin - hacyan , d. , donofrio , m. , & sulentic , j. w. 2003 , , 125 , 1897 , marcum , p. m. , aars , c. e. , fanelli , m. n. 2004 , , 127 , 3213 , melnick , j. , & sargent , w. l. w. 1977 , , 215 , 401 , mihos , j. c. , & hernquist , l. 1994 , , 425 , 13 , moore , b. , katz , n. , lake , g. , dressler , a. , & oemler , a. 1996 , nature , 379 , 613 , moss , c. , whittle , m. 2000 , , 317 , 667 , oemler , a. 1974 , , 194 , 1 , osullivan , e. , forbes , d. a. , & ponman , r. j. 2001 , , 328 , 461 , park , c. , choi , y .- y . ,
vogeley , m. s. , gott , j. r. , & blanton , m. r. 2007 , , 658 , 898 ( paper ii ) , park , c. , et al .
2005 , , 633 , 11 , park , c. , & choi , y .- y .
2005 , , 635 , l29 , pier , j. r. , munn , j. a. , hindsley , r. b. , hennessy , g. s. , kent , si . m. , lupton , r. h. , & ivezc , r. 2003 , , 125 , 1559 , postman , m. , & geller , m. 1984 , , 281 , 95 , prada , f. , gutierrez , c. m. , peletier , r. f. , & mckeith , c. d. 1996 , , 463 , 9 , ramella , m. , giuricin , g. , mardirossian , f. , & mezzetti , m. 1987 , , 188,1 , rojas , r. r. , vogeley , m. s. , hoyle , f. , & brinkmann , j. 2005 , , 624 , 571 , schechter , p. 1976
, , 203 , 297 , schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 , smith , j. a. , et al .
2002 , , 123 , 2121 , sotnikova , n. y. 1990 , in proc .
iau colloq .
124 , paired and interacting galaxies , ed sulentic j. w. , keel w. c. , & tesco c. m. , ( nasa conf .
3098 ) , 717 , sofue , y. 1994 , , 423 , 207 , stoughton , c. , et al .
2002 , , 123 , 485 , strauss , m. a. , et al .
2002 , , 124 , 1810 , tanaka , m. , et al .
2004 , , 128 , 2677 , tegmark , m. , et al .
2004 , , 606,702 , toomre , a. , & toomre , j. 1972 , , 178 , 632 , tucker , d. , et al .
2006 , an , in press , varela , j. , moles , m. , marquez , i. , galletta , g. , masegosa , j. , & bettoni , d. 2004 , , 420 , 873 , voges , w. , et al .
1999 , , 349 , 389 , wallin , j. f. , & stuart , b. v. 1992 , , 399 , 29 , weinmann , s. m. , van den bosch , f. c. , yang , x. , & mo , h. j. 2006 , , 366 , 2 , wechsler , r. h. , zentner , a. r. , bullock , j. s. , kravtsov , a. v. , & allgood , b. 2006 , , 652 , 71 , york , d. , et al .
2000 , , 120 , 1579 | we present a unified picture on the evolution of galaxy luminosity and morphology .
galaxy morphology is found to depend critically on the local environment set up by the nearest neighbor galaxy in addition to luminosity and the large scale density . when a galaxy is located
farther than the virial radius from its closest neighbor , the probability for the galaxy to have an early morphological type is an increasing function only of luminosity and the local density due to the nearest neighbor ( @xmath0 ) .
the tide produced by the nearest neighbor is thought to be responsible for the morphology transformation toward the early type at these separations . when the separation is less than the virial radius , i.e. when @xmath1 , its morphology depends also on the neighbor s morphology and the large - scale background density over a few mpc scales ( @xmath2 ) in addition to luminosity and @xmath0 .
the early type probability keeps increasing as @xmath0 increases if its neighbor is an early morphological type galaxy .
but the probability decreases as @xmath0 increases when the neighbor is a late type .
the cold gas streaming from the late type neighbor can be the reason for the morphology transformation toward late type .
the overall early - type fraction increases as @xmath2 increases when @xmath1 .
this can be attributed to the hot halo gas of the neighbor which is confined by the pressure of the ambient medium held by the background mass .
we have also found that galaxy luminosity depends on @xmath0 , and that the isolated bright galaxies are more likely to be recent merger products .
we propose a scenario that a series of morphology and luminosity transformation occur through distant interactions and mergers , which results in the morphology luminosity local density relation . |
we consider the second order difference equation defined by @xmath3 for all integers @xmath4 , with initial condition @xmath2 , where @xmath1 are real parameters .
this recurrence has recently attracted some attention , and several particular cases were already studied .
notice that the difference equation @xmath5 can be obviously reduced to ( [ 1 ] ) if @xmath6 .
as far as we know , the first particular case of ( [ 1 ] ) considered in the literature was @xmath7 , with positive initial conditions .
inar @xcite established a formula for the subsequences of even and odd terms of the solutions , respectively .
stevi @xcite gave further insight for this case , showing that every solution of ( [ 1 ] ) converges to zero if @xmath8 and @xmath9 , for all positive integer @xmath10 . if @xmath11 , then eq . (
[ 1 ] ) is @xmath12-periodic .
the case when @xmath13 , @xmath14 , and the initial conditions are nonnegative was also considered by inar @xcite , who stated a similar formula for the subsequences of even and odd terms of the solutions .
later , andruch - sobilo and migda @xcite proved that these subsequences are convergent ; moreover , it is shown that they converge to zero if @xmath15 . the case @xmath16 , @xmath14 , and arbitrary initial conditions such that @xmath17 was addressed in @xcite .
the author finds the representation formula as in the previous cases and proves that for @xmath18 , one of the subsequences converges to zero and the other one diverges .
aloqeili @xcite investigated ( [ 1 ] ) in the case when @xmath13 , @xmath19 , and proved some interesting results on the asymptotic behaviour of the solutions .
he shows that they typically converge to zero if @xmath20 and are oscillating if @xmath21 .
andruch - sobilo and migda @xcite considered the case @xmath22 , @xmath14 , with nonnegative initial conditions , showing that the subsequences of even and odd terms are monotone . in this paper , we succeed in giving a complete picture of the asymptotic behaviour of the solutions of ( [ 1 ] ) depending on the involved parameters and the initial data .
an advantage of our approach is that it allows us to consider arbitrary initial conditions , and to split our study in three cases , depending only on the values of the parameter @xmath23 .
we get as particular cases all the mentioned known results in the literature about equation ( [ 1 ] ) , but also other cases are solved here for the first time .
our results on bifurcation and stability of the solutions in sections 4 and 5 are also new . the key feature of the solutions to ( [ 1 ] )
is that the sequence @xmath24 defined by @xmath25 for all @xmath4 solves a rational difference equation of mbius type @xmath26 which is reducible to a linear difference equation ( see , e.g. , @xcite ) .
we notice that other authors refer to equation ( [ 2 ] ) as a _ ricatti difference equation _
* section 3.3 ) , ( * ? ? ?
* section 1.6 ) .
when @xmath27 , equation ( [ 2 ] ) is equivalent to the pielou logistic equation ( * ? ?
* example 3.3.4 ) , ( * ? ? ?
* example 2.39 ) .
this fact allows us to write equation ( [ 1 ] ) in the form @xmath28 where the term @xmath29 depends only on the parameter @xmath23 and the product @xmath30 for each @xmath4 .
thus , the subsequences of even and odd terms from a solution @xmath31 of ( [ 1 ] ) are given by the expressions @xmath32 notice that if these subsequences converge , say , @xmath33 , @xmath34 , then , by continuity arguments , the pair @xmath35 satisfies @xmath36 hence , either @xmath37 or @xmath38 . in particular , @xmath39 is a @xmath12-periodic solution of ( [ 1 ] ) , in such a way that the solution @xmath31 either converges to zero or to a @xmath12-periodic solution .
for this reason , the analysis of the convergence of @xmath40 and @xmath41 is an important step in our proofs .
the paper is organized as follows : in section 2 we derive the mentioned representation for the solutions of ( [ 1 ] ) . in section 3 we describe the asymptotic behaviour of the solutions ; it is divided into three subsections depending on the values of @xmath23 . in section 4
we give an interpretation of our results in terms of a bifurcation problem .
finally , we devote section 5 to analyze the stability properties of the periodic solutions of ( [ 1 ] ) .
throughout the paper , we denote @xmath30 . in this section ,
we state a representation formula for the solutions of ( [ 1 ] ) starting at any initial condition @xmath2 , except the following cases : 1 .
@xmath42 and @xmath43 , for some @xmath44 .
2 . @xmath45 and @xmath46 , for some @xmath44 .
we emphasize that , in these cases , it is not possible to construct a complete solution @xmath47 starting at @xmath48 , since at some point the denominator in ( [ 1 ] ) becomes zero .
on the other hand , it is convenient to consider the cases @xmath49 and @xmath50 separately due to their singularity ( see propositions [ 1p ] and [ 2p ] below ) .
these facts motivate us to introduce the following definitions : [ 1d ] we say that the pair @xmath51 is an admissible initial condition for ( [ 1 ] ) if either @xmath42 and @xmath52 , or @xmath45 and @xmath53 , for all @xmath44 . solutions of ( [ 1 ] ) corresponding to admissible initial conditions are called admissible solutions . an admissible solution of ( [ 1 ] ) is called a regular solution if @xmath54 and @xmath55 .
admissible solutions that are not regular are called singular solutions . in the following two propositions ,
we describe the singular solutions of ( [ 1 ] ) .
we notice that for @xmath49 all solutions are admissible if @xmath56 , while for @xmath50 all solutions are admissible .
[ 1p ] assume that @xmath49 and @xmath56 .
then , @xmath57 and @xmath58 , for all @xmath59 . * proof .
* first , assume that @xmath60 .
hence , eq . (
[ 1 ] ) reduces to @xmath61 , and it follows by induction that @xmath57 and @xmath58 , for all @xmath59 .
if @xmath62 , then @xmath63 it is easily derived by induction that @xmath64 for all @xmath59 . on the other hand , @xmath65 thus , @xmath66 , for all @xmath67 .
the proof in the case @xmath68 is completely analogous . in this case
we obtain @xmath69 , @xmath70 , for all @xmath59 .
+ [ 2p ] if @xmath50 , then the solution of ( [ 1 ] ) with initial condition @xmath51 is @xmath12-periodic .
* proof . * using ( [ 1 ] ) , we have @xmath71 the result follows by induction . + in order to get a representation for the regular solutions of ( [ 1 ] ) , we first state the explicit expression for the solutions of the mbius recurrence ( [ 2 ] ) . for a proof of the following proposition , see , e.g. , ( * ? ? ?
* example 2.39 ) .
[ p2 ] assume that @xmath72 is in the conditions of definition [ 1d ] , and @xmath73 .
then , the solution of equation ( [ 2 ] ) starting at the initial condition @xmath74 is given by @xmath75 now we are in a position to provide a representation for all admissible solutions of ( [ 1 ] ) .
[ t1 ] denote @xmath30 .
if @xmath51 is an admissible initial condition for ( [ 1 ] ) , then the corresponding solution is given by @xmath76 for all integer @xmath67 , where @xmath77 the proof of theorem [ t1 ] follows by induction from the following result : [ co1 ] if @xmath31 is an admissible solution of ( [ 1 ] ) , then @xmath78 for all @xmath4 , where @xmath29 is defined by ( [ hn ] ) .
* proof . *
first , we assume that @xmath31 is a regular admissible solution of ( [ 1 ] ) .
denote @xmath25 .
multiplying equation ( [ 1 ] ) in both sides by @xmath79 , we get @xmath80 on the other hand , @xmath81 where @xmath82 .
since @xmath24 is the solution of ( [ 2 ] ) with initial condition @xmath83 , a direct application of proposition [ p2 ] gives formula ( [ hn ] ) . for singular solutions ,
the result is also true . from proposition [ 1p ] , we can see that ( [ fh ] ) holds for @xmath49 with @xmath84 for all @xmath4
. using proposition [ 2p ] , it is clear that ( [ fh ] ) is satisfied for @xmath50 with @xmath85 for all @xmath4 .
+ the formulas given for some particular cases of equation ( [ 1 ] ) in references @xcite are particular cases of theorem [ t1 ] .
in this section , we use the representation formula given in section 2 to study the asymptotic behaviour of the solutions to ( [ 1 ] ) .
we will consider three different cases .
when @xmath16 , the expression for all admissible solutions given in theorem [ t1 ] becomes very simple : @xmath86 for all @xmath67 .
we notice that all solutions with @xmath87 are admissible .
moreover , they are regular if @xmath88 . thus , we have the following result : [ tm1 ] if @xmath16 , then all regular solutions of ( [ 1 ] ) are unbounded .
moreover , if @xmath31 is a regular solution of ( [ 1 ] ) , then some subsequences of @xmath31 are divergent and other converge to zero .
the singular solutions are @xmath12-periodic if @xmath89 , and @xmath90-periodic if @xmath49 .
* proof . *
it follows easily from the relations ( [ m1])-([m12 ] ) .
the complete behaviour of the solutions is the following ( here @xmath91 if @xmath92 , and @xmath93 if @xmath94 ) : 1 . if @xmath95 then @xmath96 , @xmath97 2 . if @xmath89 then @xmath98 , @xmath99 for all @xmath67 .
3 . if @xmath100 then @xmath101 , @xmath102 4 .
if @xmath103 then @xmath101 , @xmath104 and @xmath105 5 . if @xmath49 then @xmath106 , @xmath107 , @xmath108 , @xmath109 for all @xmath67 .
thus , @xmath110 , for all @xmath111 .
6 . if @xmath112 then @xmath113 , @xmath114 and @xmath97 the main result in this section is the following : [ tam1 ] if @xmath117 and @xmath116 , then all admissible solutions of ( [ 1 ] ) converge to zero if @xmath55 , and are @xmath12-periodic if @xmath50 .
* proof . * from proposition [ 2p ] , we already know that the solutions of ( [ 1 ] ) are @xmath12-periodic if @xmath50 .
next we assume that @xmath55 .
we first address the case @xmath42 .
as we already mentioned in the introduction , it was proved in @xcite that all regular solutions of ( [ 1 ] ) converge to zero if @xmath7 .
the same arguments of theorem 1 in @xcite apply to the case @xmath42 , @xmath118 , using theorem [ t1 ] .
it remains the case @xmath119 , @xmath55 .
let @xmath31 be an admissible solution of ( [ 1 ] ) .
using proposition [ co1 ] , and the fact that @xmath120 , we have : @xmath121 the dalembert criterion ensures that @xmath122 .
hence , @xmath123 .
we begin this subsection with a simple result corresponding to the case @xmath125 .
notice that in this case all solutions are admissible if @xmath54 .
[ p0 ] if @xmath125 , then all admissible solutions of ( [ 1 ] ) are @xmath12-periodic .
* proof . * in this case , eq .
( [ 1 ] ) becomes @xmath126 thus , @xmath127 for all @xmath128 .
+ the proof of the following lemma is very easy from expression ( [ hn ] ) , so we omit it : [ lemah ] let @xmath45 and @xmath53 , for all @xmath44 .
then , @xmath129 , where @xmath130 for all @xmath4 .
in order to address the case @xmath131 , we investigate the character of the subsequences of even and odd terms , which depend on the sequence @xmath132 .
for example , if @xmath133 for all sufficiently large @xmath134 , then it is clear from ( [ fh ] ) that the subsequence of odd terms is eventually increasing .
notice that , in view of lemma [ lemah ] , @xmath135 if and only if @xmath136 , and @xmath137 if and only if @xmath138 [ ph ] assume that @xmath131 , @xmath54 , @xmath55 , and @xmath53 , for all @xmath44
. then , there exists @xmath139 such that the sequences @xmath140 and @xmath141 have constant sign for all @xmath142
. * proof .
* we first consider the case @xmath143 , and distinguish three situations : 1 . if @xmath143 and @xmath144 , then @xmath145 and @xmath146 for all @xmath4 .
thus , @xmath136 for all @xmath147 .
2 . if @xmath143 and @xmath148 , then @xmath149 and @xmath150 for all @xmath4 . thus , @xmath151 for all @xmath147 .
3 . if @xmath143 and @xmath152 , then @xmath149 for all @xmath4 . on the other hand , since @xmath153 we can conclude that there exists @xmath154 such that @xmath136 for all @xmath155 .
analogously , we consider the same situations for @xmath156 . 1 . if @xmath157 and @xmath144 , then @xmath158 and @xmath159 for all @xmath67 .
since @xmath146 for all @xmath4 , it follows that @xmath160 and @xmath161 for all @xmath67 .
2 . if @xmath156 and @xmath148 , then @xmath162 and @xmath163 for all @xmath67 .
since @xmath164 it follows that there exists @xmath165 such that @xmath166 and @xmath167 for all @xmath168 .
3 . if @xmath156 and @xmath152 , then @xmath162 and @xmath163 for all @xmath67 .
since @xmath153 it follows that there exists @xmath169 such that @xmath160 and @xmath161 for all @xmath170 . as a consequence of proposition [ ph ] , we have : [ cormon ] if @xmath131 , and @xmath31 is a regular solution of ( [ 1 ] ) , then the subsequences @xmath40 and @xmath41 are eventually monotone . *
proof . * from proposition [ ph ] , we know that there exists @xmath139 such that the sequences @xmath140 and @xmath141 have constant sign for all @xmath142 .
assume that @xmath160 for all @xmath142 .
then , @xmath171 for all @xmath142 . on the other hand ,
since @xmath172 , it is clear that there exists @xmath173 such that @xmath174 for all @xmath175 . since , by proposition [ co1 ] , @xmath176 , it follows that @xmath177 , that is , the subsequence @xmath178 is decreasing .
the remainder cases are analogous .
+ using this corollary , we can prove the following key result : [ pconv ] if @xmath131 , and @xmath31 is a regular solution of ( [ 1 ] ) , then the subsequences @xmath40 and @xmath41 are convergent
. * proof .
* we only prove this result for the sequences of even terms , since the other case is completely analogous . as we noticed above , @xmath172 , and therefore @xmath179 for all sufficiently large @xmath10 .
without loss of generality , we assume that @xmath179 for all @xmath4 .
since @xmath40 is eventually monotone , we only have to prove that it is bounded . for it
, we use theorem [ t1 ] and lemma [ lemah ] : @xmath180 for the inequality above , we have used that @xmath181 for all @xmath182 . since @xmath183 it follows from the dalembert rule that the series @xmath184 is convergent .
this ensures that @xmath40 is bounded .
+ finally , we can state the main result of this subsection for the regular solutions of ( [ 1 ] ) .
[ tap1 ] if @xmath131 , then all regular solutions of ( [ 1 ] ) converge to a @xmath12-periodic solution @xmath35 of ( [ 1 ] ) , with @xmath185 . * proof .
* by proposition [ pconv ] , there exist @xmath186 , @xmath187 . as it was mentioned in the introduction ,
the sequence @xmath188 is a solution of ( [ 2 ] ) and , by proposition [ p2 ] , @xmath189 taking limits as @xmath190 in ( [ 1 ] ) , it is clear that the relations ( [ pq ] ) hold , and therefore @xmath35 is a @xmath12-periodic solution of ( [ 1 ] ) .
+ using theorem [ tap1 ] and propositions [ 1p ] , [ 2p ] and [ p0 ] , we can describe completely the behaviour of all admissible solutions in the case @xmath124 .
[ tap2 ] assume that @xmath124 , and @xmath31 is an admissible solution of ( [ 1 ] ) .
then : 1 . if either @xmath125 or @xmath191 , then @xmath31 is @xmath12-periodic
if @xmath56 , @xmath54 and @xmath55 , then @xmath31 converges to a @xmath12-periodic solution .
3 . if @xmath56 , @xmath49 and @xmath192 , then @xmath31 is unbounded .
4 . if @xmath193 then @xmath194 for all @xmath44 .
[ r311 ] notice that zero is the unique equilibrium of ( [ 1 ] ) if @xmath195 . when @xmath196 , there are two nontrivial equilibrium points @xmath197 .
thus , the minimal period of the @xmath12-periodic solution @xmath35 mentioned in theorem [ tap2 ] is actually @xmath198 if @xmath196 and @xmath199 . as a by - product of theorems [ tm1 ] , [ tam1 ] and [ tap2 ] , we have the following result on the boundedness of the solutions to ( [ 1 ] ) : [ pbound ] all admissible solutions of ( [ 1 ] ) are bounded , except in the following two cases : 1 .
@xmath16 and @xmath200 ; 2 .
@xmath124 , @xmath49 , and @xmath192 .
the analysis made in section 3 can be also viewed in terms of bifurcation diagrams .
first , notice that the case @xmath14 may be reduced to @xmath201 by the change of variables @xmath202 , and the case @xmath203 may be reduced to @xmath19 by the change of variables @xmath204 .
thus , we can view equation ( [ 1 ] ) as a one - parameter family of difference equations depending only on @xmath23 , if we consider the cases @xmath14 , @xmath60 , and @xmath203 separately . as an example , we consider the case @xmath19 .
there are two _ regular _ bifurcation points in @xmath16 and @xmath42 .
when @xmath20 , all regular solutions converge to zero ; as @xmath23 passes through @xmath198 to the left , the @xmath205-limit set of any regular solution is a @xmath12-periodic point .
one of the branches of this periodic solution in the bifurcation diagram approaches zero as @xmath23 tends to @xmath206 , and the other one diverges to @xmath207 or @xmath208 . after crossing the other bifurcation point @xmath16 , only the bounded branch remains , and all regular solutions are attracted by zero . ) with @xmath19 , @xmath209 , @xmath210 . ]
if we plot the bifurcation diagram corresponding to an initial condition @xmath51 with @xmath54 , we also observe a _ singular _
bifurcation point when @xmath211 ( that is , for @xmath212 ) if @xmath213 .
indeed , for all values of @xmath23 in a neighbourhood of @xmath214 the limit of the solution is zero , while for @xmath215 the solution is @xmath12-periodic . in figure
[ bif1 ] , we plotted the bifurcation diagram corresponding to @xmath19 and the initial condition @xmath216 .
we observe the singular bifurcation point @xmath217 , for which the solution is @xmath12-periodic . + as it may be seen from proposition
[ pbound ] , the case when @xmath89 is special because all admissible solutions of ( [ 1 ] ) are bounded . for @xmath19 ,
this happens when @xmath218 .
we plot in figure [ bif2 ] the bifurcation diagram corresponding to the initial data @xmath219 . on the right we plotted a magnification for @xmath23 close to @xmath206 in order to emphasize that the branches of periodic points for this initial condition are continuous on the right at @xmath16 .
of course , there is a discontinuity on the left , since for @xmath16 the solution is @xmath12-periodic , and for @xmath220 it converges to zero . ) with @xmath19 , @xmath209 , @xmath221.,title="fig : " ] ) with @xmath19 , @xmath209 , @xmath221.,title="fig : " ] we produced the bifurcation diagrams in a standard way : for all values of @xmath23 with a step @xmath222 , we constructed the first @xmath223 iterations corresponding to the initial conditions @xmath216 and @xmath219 , and plotted them starting at @xmath224 . for the magnification in figure [ bif2 ]
, we produced @xmath225 iterations for each value of @xmath23 , with a step @xmath226 .
as we have shown , the @xmath205-limit set of a bounded solution of ( [ 1 ] ) is a periodic solution of minimal period @xmath198 , @xmath12 or @xmath90 . in this section ,
we study the stability properties of these periodic solutions .
we begin with the zero solution . the zero solution of ( [ 1 ] )
is asymptotically stable if and only if either @xmath119 or @xmath42 and @xmath118 .
moreover , in both cases it attracts all regular solutions .
* proof . * the characteristic equation associated to the linearization of ( [ 1 ] ) at the equilibrium @xmath227 is given by the quadratic equation @xmath228 .
thus , the zero solution is locally asymptotically stable if @xmath119 , and unstable if @xmath124 .
theorem [ tam1 ] shows that zero is actually a global attractor of all regular solutions when @xmath229 . if @xmath42 and @xmath60 , then all solutions are @xmath12-periodic ( the minimal period may be one ) , and they are clearly stable but not asymptotically stable . finally , if @xmath16 , it follows from theorem [ tm1 ] that zero is unstable , since solutions starting at initial conditions arbitrarily close to @xmath230 are unbounded .
+ next we deal with the nontrivial periodic solutions . the unique periodic solutions with period greater than @xmath12 are the @xmath90-periodic points indicated in theorem [ tm1 ] for @xmath16 and @xmath49 .
they are clearly unstable .
as proved in proposition [ p0 ] , all admissible solutions of ( [ 1 ] ) for @xmath125 are @xmath12-periodic ; moreover , from the proof of this proposition , it is clear that they are stable .
if @xmath56 , then the @xmath12-periodic points of ( [ 1 ] ) are defined by the initial conditions @xmath35 such that @xmath231 .
this can be easily seen taking into account the correspondence between the solutions of ( [ 1 ] ) and the orbits of the discrete dynamical system associated to the map @xmath232 defined by @xmath233 the @xmath12-periodic solutions of ( [ 1 ] ) are defined by the fixed points of the map @xmath234 .
it is straightforward to prove that , if @xmath56 and @xmath235 , @xmath236 if and only if @xmath237 .
notice that , as mentioned in remark [ r311 ] , equation ( [ 1 ] ) has two nontrivial equilibria @xmath197 if @xmath238 ; thus , the minimal period of @xmath35 is one if @xmath239 . otherwise , the minimal period is two .
direct computations show that the linearization of @xmath240 at any point @xmath35 satisfying @xmath231 has two eigenvalues : @xmath23 and @xmath198 .
thus , all nonzero @xmath12-periodic solutions are unstable for @xmath119 .
when @xmath16 or @xmath42 , it follows from theorems [ tm1 ] , [ tam1 ] that they are also unstable . since we already studied the case @xmath125 , the remainder part of this section is devoted to prove that every nonzero periodic solution of ( [ 1 ] ) is stable if @xmath131 .
for it , we will use the formula given in theorem [ t1 ] .
first , we need some bounds for the involved products .
[ ln1 ] assume that @xmath131 and @xmath241 .
for all @xmath4 one has : @xmath242 and , as a consequence , @xmath243 * proof . * since @xmath241
, we have @xmath244 [ pn1 ] if @xmath131 and @xmath241 , then 1 .
@xmath245 2 .
@xmath246 * proof . * 1 . as in the proof of proposition [ pconv ] , one gets @xmath247 first , if @xmath248 then @xmath249 and the result is trivial .
+ if @xmath250 and @xmath13 we have that @xmath251 and hence @xmath252 therefore , @xmath253 + on the other hand , if @xmath250 and @xmath22 then @xmath254 and this implies @xmath255 thus , @xmath256 2 .
the same argument used in the proof of proposition [ pconv ] shows @xmath257 + again , if @xmath258 then @xmath259 and the inequality of the statement is straightforward .
+ when we suppose that @xmath260 and @xmath13 then , as we did before , @xmath261 and , hence , @xmath262 + finally , if @xmath260 and @xmath22 , then @xmath263 , and therefore @xmath264 since @xmath241 , we get that @xmath265 which leads us to @xmath266 therefore , @xmath267
+ [ pn2 ] if @xmath131 and @xmath241 , then 1 .
@xmath268 2 .
@xmath269 * proof . *
first , notice that we have @xmath270 if @xmath271 then @xmath272 and when @xmath273 we get @xmath274 therefore , we get the inequality @xmath275 and the result claimed follows at once from proposition [ pn1 ] . + if @xmath131 then every nonzero periodic solution of ( 1.1 ) is stable .
* proof . * as mentioned above , if @xmath131 then every nonzero periodic solution @xmath31 of ( 1.1 ) is given by @xmath276 for all @xmath277 where @xmath278 .
let us , then , fix @xmath279 such that @xmath280 since the mapping @xmath281 is continuous , we may find @xmath282 such that @xmath283 whenever @xmath284 .
let @xmath31 be the solution of ( 1.1 ) obtained for some initial conditions @xmath51 verifying @xmath284 . according to propositions
[ pn1 ] and [ pn2 ] , there exist continuous functions @xmath285 such that @xmath286 for @xmath287 and @xmath288 for all @xmath67 . for every @xmath289 , we can therefore find @xmath290 such that @xmath291 implies @xmath292 where @xmath293 .
this clearly implies that , for every @xmath67 , @xmath294 since , by theorem [ t1 ] , @xmath295 it follows that @xmath296 now , if we choose @xmath297 and @xmath298 then we have @xmath299 which implies @xmath300 and then @xmath301 the same argument applied to the subsequence @xmath41 completes the proof . +
we described completely the dynamics and stability properties of equation ( [ 1 ] ) for all values of the real coefficients @xmath1 .
this was possible because of the relation of this equation with the mbius recurrence ( [ 2 ] ) .
we list some open problems related to eq .
( [ 1 ] ) . 1 .
analyze the behaviour of the solutions of ( [ 1 ] ) when the coefficients and the initial conditions are complex .
for a recent work on the dynamics of mbius transformations with complex coefficients , see @xcite .
2 . in the case
@xmath124 , try to determine the actual value of the @xmath12-periodic point @xmath35 to which a regular solution of ( [ 1 ] ) converges for a given initial condition @xmath51 . in the light of theorem [ t1 ] , this is equivalent to find the value of the infinite products @xmath302 , @xmath303
this research was supported in part by the spanish ministry of science and innovation ( formerly _ ministry of science and education _ ) and feder , grant mtm2007 - 60679 .
10 r. p. agarwal , _ difference equations and inequalities .
theory , methods , and applications , _ second edition , monographs and textbooks in pure and applied mathematics , * 228 * , marcel dekker , inc .
, new york , 2000 .
m. aloqeili , dynamics of a rational difference equation , appl .
comput . * 176 * ( 2006 ) , 768774 .
a. andruch - sobilo and m. migda , further properties of the rational recursive sequence @xmath304 , opuscula math .
* 26 * ( 2006 ) , 387394 .
a. andruch - sobilo and m. migda , on the rational recursive sequence @xmath305 , preprint .
a. cima , a. gasull , and v. maosa , dynamics of some rational discrete dynamical systems via invariants , internat .
chaos appl .
* 16 * ( 2006 ) , 631645 . c. inar , on the positive solutions of the difference equation @xmath306 , appl .
* 150 * ( 2004 ) , 2124 . c. inar , on the positive solutions of the difference equation @xmath307 , appl
* 156 * ( 2004 ) , 587590 . c. inar , on the solutions of the difference equation @xmath308 , appl .
* 158 * ( 2004 ) , 793797 . c. inar , on the positive solutions of the difference equation @xmath309 , appl
* 158 * ( 2004 ) , 809812 . c. inar , on the difference equation @xmath310 , appl .
comput . * 158 * ( 2004 ) , 813816 .
p. cull , m. flahive and r. robson , _ difference equations . from rabbits to chaos , _ springer , new york , 2005 .
s. elaydi , _ an introduction to difference equations , _ third edition , springer , new york , 2005 .
m. r. s. kulenovi and g. ladas , _ dynamics of second order rational difference equations , _ chapman & hall / crc , boca raton , 2002 .
s. stevi , more on a rational recurrence relation , appl .
e - notes * 4 * ( 2004 ) , 8085 . | we describe the asymptotic behaviour and the stability properties of the solutions to the nonlinear second order difference equation @xmath0 for all values of the real parameters @xmath1 , and any initial condition @xmath2 . and second order difference equation , ricatti difference equation , periodic solution , stability , asymptotic behaviour |
the on a generic smooth surface @xmath3 embedded in three - dimensional euclidean space consists of the points where @xmath3 has zero gaussian curvature .
it separates ( where the curvature is positive ) from ( where the curvature is negative ) .
these notions are well - defined for surfaces embedded in affine or even projective space , as the sign of gaussian curvature is invariant under affine transformations . if the surface @xmath3 is expressed locally as the graph @xmath4 of a smooth function @xmath5 , then the sign of its hessian determinant @xmath6 equals the sign of its curvature at the corresponding point .
thus the parabolic curve is the image under @xmath5 of its , which is defined by @xmath7 .
when the surface @xmath3 is the graph of a polynomial @xmath8 $ ] , this local description is global , and so questions about the disposition of the parabolic curve on @xmath3 are equivalent to the same questions about the hessian curve in @xmath9 .
suppose that @xmath10 is even .
harnack proved @xcite that a smooth plane curve of degree @xmath10 has at most @xmath11 connected components in @xmath12 .
this is also the bound for the number of components of a compact curve in @xmath9 of degree @xmath10 .
a non - compact curve in @xmath9 of degree @xmath10 can have at most @xmath13 bounded components ( ) and @xmath10 unbounded components .
these unbounded components come from the intersection of the corresponding curve in @xmath12 with the line at infinity .
harnack constructed a curve in @xmath12 of degree @xmath10 with @xmath11 components which has one component meeting the line at infinity in @xmath10 points .
this harnack curve shows that the bound for non - compact curves in @xmath9 is attained , and choosing a different line at infinity shows that the bound for compact curves in @xmath9 is also attained .
we are interested in the possible number and disposition of the components of the hessian curve in @xmath9 of a polynomial @xmath14 $ ] of degree @xmath15 .
this is problem 2001 - 1 in the list of arnold s problems @xcite , attributed to a. ortiz - rodrguez .
see also the discussion of related problems 2000 - 1 , 2000 - 2 , 2001 - 1 , and 2002 - 1 .
the hessian of @xmath5 has degree at most @xmath16 . by harnack s theorem
, a compact hessian curve has at most @xmath17 ovals and a non - compact hessian curve has at most @xmath18 ovals and @xmath19 unbounded components .
while we know of no additional restrictions on hessian curves , we are not assured that all possible configurations are acheived by hessians .
when @xmath15 is at least 4 , simple parameter counting shows that not all polynomials of degree @xmath19 arise as hessians of polynomials of degree @xmath15 .
the placement of the set of hessian curves among all curves of degree @xmath19 may restrict the possible configurations of hessian curves in @xmath9 .
for example , a simple calculation shows that @xmath20 thus the hessian of a polynomial is a linear combination of 3 squares , which shows that the hessians lie in the second secant variety to the veronese embedding of polynomials of degree @xmath21 in polynomials of degree @xmath22 ( the veronese consists of the perfect squares ) .
we also know of no general techniques for constructing hessian curves with a prescribed configuration .
one of us ( ortiz - rodrguez ) investigated this question @xcite and constructed polynomials @xmath23 $ ] of degree @xmath15 whose hessians had @xmath24 ovals in @xmath9 . when @xmath15 is 4 , 5 , and 6 , these numbers are 3 , 6 , and 10 , respectively .
we do not know if it is possible for a hessian curve to achieve the harnack bound , or more generally , which configurations are possible for hessian curves . here , we present a quartic polynomial @xmath5 whose hessian achieves the harnack bound of 4 ovals , a quintic whose hessian has 8 ovals , a sextic whose hessian has 11 ovals , as well as examples of non - compact hessian curves of quartics , quintics , and sextics .
these examples show that hessian curves can have more ovals than was previously known .
they were found in a computer search , using the software maple . our method was to generate a random polynomial ,
compute its hessian , and then compute an upper bound on its number of ovals in @xmath12 , sometimes also screening for the number of unbounded components in @xmath9 .
this upper bound was one - half the minimum number of real critical points of a projection to one of the axes , as each oval in @xmath12 contributes at least two critical points to the projection .
we separately investigated compact hessian curves of sextics .
polynomials whose upper bound for ovals was at least 4 , 8 , and 11 ( for quartics , quintics , and sextics , respectively ) were saved for further study .
the further investigation largely involved viewing pictures in @xmath9 of these potentially interesting hessians . in all , only a few hundred polynomials warranted such further scrutiny .
we examined the hessians of 150 million quartics , 40 million each of quintics and sextics , and over 200 million sextics with compact hessians ( the different protocol of pre - screening for compactness allowed a greater number to be examined ) .
this required 628 days of cpu time on several computers , most of which were running linux on intel pentium processors with speeds between 1.8 and 3 gigahertz .
we did not find a quartic whose hessian had 3 ovals and 4 unbounded components , nor a quintic whose hessian had more than 8 ovals in @xmath12 , nor a sextic whose hessian had more than 11 ovals in @xmath12 .
( the examples we give at the end with 12 ovals in @xmath12 are pertubations of a curve we found with 11 ovals . )
this suggests that it may not be possible for hessian curves in @xmath9 to achieve the harnack bounds .
further pictures and computer code are at the web page
. tables [ t1 ] and [ t2 ] summarize this discussion concerning the number of components of hessian curves .
the pairs @xmath25 in table [ t2 ] refer to ovals and unbounded components , respectively .
each oval has exactly two vertical and two horizontal tangents , and each unbounded component has exactly one vertical and one horizontal tangent .
these claims are best verified symbolically . for each , we give the polynomial @xmath5 and display a picture of the hessian curve , drawn with the implicitplot function of maple .
these were rendered , at least locally , with a grid size sufficiently small to separate the tangents , and therefore provide a faithful picture of the hessian curves as curves in @xmath9 .
figure [ f : two](a ) displays the hessian curve of the quartic @xmath26 which has 3 ovals and 2 unbounded components .
figure [ f : two](b ) displays the hessian curve of the quartic @xmath27 which has 2 ovals and 4 unbounded components .
despite examining over 240 million sextics , we did not find any sextics whose hessian curves had more than 11 ovals in @xmath12 .
these last two examples , which have 12 ovals in @xmath12 , are perturbations of a sextic found in the search whose hessian curve had 11 ovals in @xmath12 . | we give some real polynomials in two variables of degrees @xmath0 , @xmath1 , and @xmath2 whose hessian curves have more connected components than had been known previously .
in particular , we give a quartic polynomial whose hessian curve has 4 compact connected components ( ovals ) , a quintic whose hessian curve has 8 ovals , and a sextic whose hessian curve has 11 ovals . |
let @xmath10 be the surface with genus @xmath11 and @xmath12 marked points . the _ mapping class group _
@xmath13 is the group @xmath14 where @xmath15 is the group of orientation preserving self - homeomorphisms of the surface @xmath10 which permute the marked points and @xmath16 is the path component of @xmath17 containing the identity map .
let @xmath18 be the set of @xmath19 marked points .
an _ essential curve _
@xmath20 in @xmath10 is a simple closed curve in @xmath21 such that any disk in @xmath10 bounded by @xmath20 must contain at least two marked points .
the isotopy class of @xmath20 is its orbit under the action of @xmath16 .
a _ curve system _ is a non - empty set of isotopy classes of essential curves which have pairwise disjoint representatives and a _ sub curve system _ of a curve system is any non - empty subset of a curve system .
the _ curve complex _
@xmath22 is the simplicial complex whose vertices correspond to isotopy classes of essential curves in @xmath10 and whose @xmath23-simplices correspond to curve systems with @xmath24 curves .
henceforth we will abuse notation by not distinguishing between essential curves and their isotopy classes . in the cases of the sporadic surfaces of low complexity , @xmath25 , @xmath26 and @xmath27 ,
it is customary to include simplices for sets of essential curves with pairwise minimal intersection , but for group cohomological reasons we will not do this here .
hence for our purposes @xmath28 , @xmath29 and @xmath30 will be disconnected complexes of dimension @xmath31 .
the curve complex plays dual geometric and homological roles for the structure of the mapping class group . on the one hand
we have results in the vein of ivanov @xcite , which retrieve the mapping class group as the set of simplicial isomorphisms of the curve complex and other related simplicial complexes .
on the other hand we have the work of harer @xcite which shows that the single non - trivial reduced homology group of the curve complex is the dualizing module for the mapping class group , thereby linking its group homology and cohomology .
one should not be too surprised at these dual functions for the curve complex , as the curve complex was originally proposed by harvey @xcite as an analog for the mapping class group of the spherical building at infinity for nonuniform lattices in semi - simple lie groups .
recent work of aramayona and leininger @xcite advanced the ivanov side of the picture by giving a finite rigid subcomplex @xmath0 in the curve complex with the property that any simplicial injection of @xmath0 into the curve complex is induced by a unique mapping class . in this paper
we relate the work in @xcite to the harer side of that picture by relating the finite rigid set @xmath0 of aramayona and leininger to a homologically non - trivial sphere @xmath9 in the curve complex whose orbit under the action of the mapping class group generates the reduced homology of the curve complex .
first , we focus on the sphere with @xmath19 marked points . in proposition
[ prop : simpsphere ] we find , for each @xmath5 , an explicit set of curves on @xmath32 whose associated vertices in @xmath33 determine an essential simplicial sphere in @xmath33 .
the information in proposition [ prop : simpsphere ] will allow us to prove our first main result in @xmath34[sec : al ] : [ thm : same ] assume that @xmath5 .
then the finite rigid set @xmath35 given in @xcite is precisely the essential @xmath36-sphere @xmath37 whose reduced homology class is determined in proposition [ prop : simpsphere ] of this paper to be a @xmath38-module generator of the reduced homology of the curve complex .
this answers question 2 of @xcite in the affirmative for @xmath5 .
given the surprising coincidence of our essential sphere and the finite rigid set of aramayona and leininger for the surface @xmath39 , and the fact that @xmath40 , one might be tempted to conjecture that homologically non - trivial spheres in the curve complex are always finite rigid sets . attempting to test our conjecture in the cases
@xmath41 we learned that it is not true . in proposition
[ prop : geng ] we construct essential spheres @xmath42 and @xmath43 in @xmath44 and @xmath45 respectively .
our second main result , proved in @xmath34[sec : gcomp ] is : [ thm : notiff ] let @xmath46 and @xmath47 or @xmath48 and @xmath49 .
then the essential @xmath50-sphere @xmath51 that is determined in proposition [ prop : geng ] ( i ) represents a @xmath38-module generator for the reduced homology of @xmath52 , ( ii ) is a proper subset of @xmath53 , but ( iii ) is not rigid . modified versions of theorems [ thm : same ] and [ thm : notiff ]
hold for the sporadic surfaces @xmath25 , @xmath26 and @xmath27 .
see @xmath34[ss : sporadic ] for the precise statements .
while theorem [ thm : notiff ] rules out the possibility that homologically non - trivial spheres in the curve complex must be rigid , we have the following suggestive corollary : suppose @xmath54 and @xmath55 or @xmath56 and @xmath47 .
let @xmath57 be the finite rigid set given in @xcite .
then the inclusion map @xmath58 induces a homomorphism @xmath59 with non - trivial image . in both cases
the inclusion map @xmath60 factors through the inclusion map @xmath61 and induces a homomorphism @xmath62 with non - trivial image .
suppose @xmath63 .
does the inclusion map @xmath64 of the finite rigid set given in @xcite induce a homomorphism @xmath59 with non - trivial image ? here is a guide to the paper .
we begin , in @xmath34[sec : review of the background ] , with a review of the necessary background .
the associahedron will enter into this review because it concisely describes the simplicial structure of the spheres in the curve complex which we demonstrate to be homologically non - trivial .
we introduce it in @xmath34[sec : back ] .
our new work begins in @xmath34[sec : surfaces of genus 0 ] , where we restrict our discussion to surfaces of genus 0 with @xmath65 marked points and give a finite @xmath66-module resolution of the steinberg module ( the reduced homology of the curve complex ) using the the cellular chain complex of the arc complex relative to the arc complex at infinity .
the last two terms of this resolution give a @xmath66-module presentation of this homology group which we use to give a single class whose orbit under the mapping class group generates the entire relative homology group . in @xmath34[sec : ccsph ] we use harer s homotopy equivalence discussed in @xmath34[sec : simpmap ] to convert our non - trivial class in the homology of the arc complex relative to the arc complex at infinity to a class in the homology of the curve complex .
we then simplify our representative of this class . in proposition [ prop : simpsphere ]
we give , explicitly , a finite subset @xmath67 of the curve complex @xmath68 the orbit of whose homology class under the action of the mapping class group generates the reduced homology of the curve complex . in @xmath34[sec : al ] we find that the simplified representative is precisely the finite rigid set of aramayona and leininger , proving theorem [ thm : same ] . in @xmath34[sec :
genusg ] we consider the same situation when @xmath69 and @xmath70 .
this situation was already treated in @xcite , however we are now able to simplify the results given there drastically .
we give , in proposition [ prop : geng ] , an explicit simplified homologically non - trivial sphere in the curve complex the orbit of whose homology class under the action of the mapping class group generates the reduced homology of the curve complex .
we establish that it is a proper subset of the finite rigid set of aramayona and leininger . in @xmath34[sec : gcomp ] we prove theorem [ thm : notiff ] , by showing that it is not a finite rigid set .
modified versions of theorems [ thm : same ] and [ thm : notiff ] , for sporadic surfaces , are discussed and proved in @xmath34[ss : sporadic ] .
* acknowledgement : * we gratefully acknowledge the help of the anonymous referee , who suggested significant simplifications to the work in @xmath34[sec : ggen ] and @xmath34[sec : gcomp ] .
our results in @xmath34[sec : surfaces of genus 0 ] and @xmath34[sec : genusg ] will concern the surfaces @xmath39 where @xmath71 and @xmath10 where @xmath72 and @xmath73 .
however , many of the results that we review in this section apply in greater generality so we provide the stronger statements and more general definitions when possible .
let @xmath71 or @xmath72 and set @xmath74 by the work of harer we have the following theorem : [ fact : cc ] assume @xmath71 or @xmath72 .
the curve complex @xmath75 has the homotopy type of a countably infinite wedge sum of spheres of dimension @xmath76 thus @xmath77 where @xmath76 is as in ( [ eq : tau ] ) .
in particular the reduced homology @xmath78 of the curve complex is non - trivial only in dimension @xmath76 where we get @xmath79 the _ dualizing module _ of a duality group @xmath80 of cohomological dimension @xmath81 is a @xmath80-module @xmath82 such that for any @xmath80-module @xmath83 and any @xmath84 we have @xmath85 where @xmath86 has the diagonal module structure .
( see ( * ? ? ? *
viii.10 ) for a general introduction to duality groups . ) in particular if @xmath80 is any torsion free finite index subgroup of the mapping class group @xmath13 where @xmath71 or @xmath72 then by the work of harer ( * ? ? ?
* theorem 4.1 ) @xmath80 is a duality group with dualizing module the _ steinberg module _ @xmath87 we view this as motivation for a careful study of the homotopy type of the curve complex . to help us in this study
, it will be convenient to introduce a closely related complex , the arc complex " of a surface with marked points .
let @xmath10 be the surface with genus @xmath11 and @xmath88 marked points and let @xmath18 be the set of marked points .
an _ arc _ in @xmath10 is either the unoriented image of a properly embedded path in @xmath10 joining two points of @xmath89 which is disjoint from @xmath89 except at its endpoints or else the image of an unoriented simple loop in @xmath10 based at a point in @xmath89 disjoint from @xmath89 except at its basepoint .
essential arc _ in @xmath10 is an arc which does not bound an embedded disk in @xmath10 whose interior is disjoint from @xmath89 .
we will say that two arcs are _ disjoint _ if they are disjoint except possibly at their endpoints .
the isotopy class of an essential arc is its orbit under the elements of @xmath16 which fix the marked points pointwise .
an _ arc system _ is a set of isotopy classes of arcs with pairwise disjoint representatives and a _ sub arc system _ of an arc system is any non - empty subset of the arc system .
the _ arc complex _
@xmath90 is the simplicial complex whose vertices are isotopy classes of essential arcs and whose @xmath23-simplices correspond to arc systems with @xmath24 isotopy classes of arcs .
observe that a maximal arc system gives a triangulation of the surface @xmath10 with @xmath19 vertices so by euler characteristic it will have @xmath91 arcs .
therefore we have @xmath92 the arc complex has a very nice property : [ fact : con ] if @xmath93 or if both @xmath72 and @xmath88 then @xmath94 is contractible .
an arc system _ fills _ the surface @xmath10 if the arcs in the arc system cut the surface into disks with at most one marked point in their interior . the _ arc complex at infinity _
@xmath95 is the union of all simplices of @xmath94 whose corresponding arc systems do _ not _ fill the surface .
one may also alternatively describe @xmath96 as the subset of points in @xmath94 with infinite stabilizers under the action of @xmath13 .
now suppose that @xmath54 and @xmath71 or @xmath72 and @xmath97 and let @xmath76 be as in equation ( [ eq : tau ] ) .
at least @xmath98 pairwise disjoint arcs are needed to fill the surface @xmath10 .
hence @xmath96 contains the entire @xmath76-skeleton of @xmath94 .
if on the other hand we have @xmath72 and @xmath93 then at least @xmath99 pairwise disjoint arcs are needed to fill the surface which implies that @xmath96 contains the @xmath100-skeleton of @xmath94 but not the entire @xmath76-skeleton of @xmath94 .
this fact renders an essential part of our argument given in @xmath34[sec : res ] applicable only to the cases where @xmath54 and @xmath71 or @xmath72 and @xmath97 .
we now introduce a simplicial complex which we will later embed in the curve complex as a homologically non - trivial sphere .
let @xmath101 .
the associahedron @xmath102 is a convex polytope homeomorphic to the closed @xmath103-ball .
its boundary @xmath104 is a cell complex homeomorphic to the @xmath105-sphere .
the dual of the cell complex @xmath104 is a _ simplicial _
complex @xmath106 which is again homeomorphic to the @xmath105-sphere . here
we provide a convenient description of @xmath106 taken from ( * ? ? ?
* corollary 2.7 ) .
let @xmath107 be a graph .
a _ tube _ is any proper non - empty subset of the vertices of @xmath107 whose induced subgraph in @xmath107 is connected .
let @xmath108 and @xmath109 be two tubes of @xmath107 .
we say that @xmath108 and @xmath109 are _ nested _ if @xmath110 or @xmath111 .
the tubes @xmath108 and @xmath109 _ overlap _ if @xmath112 and @xmath108 and @xmath109 are not nested .
tubes @xmath108 and @xmath109 are _ adjacent _ if @xmath113 and @xmath114 is a tube .
two tubes in @xmath107 are _ compatible _ if they do not overlap and are not adjacent .
let @xmath115 be the simplicial complex with a vertex for each tube of @xmath107 and a @xmath23-simplex for each set of @xmath24 pairwise compatible tubes in @xmath107 .
[ thm : assoc ] the dual @xmath106 of the boundary @xmath105-sphere of the associahedron @xmath102 is the simplicial complex @xmath116 where @xmath117 is the path graph with @xmath118 vertices given in figure [ fig : pathonly ] . with vertex set @xmath119.[fig : pathonly],title="fig : " ]
( -191,11)@xmath120 ( -154.5,11)@xmath121 ( -118,11)@xmath122 ( -90,-1)@xmath123 ( -53,11)@xmath124 ( -17,11)@xmath118 @xmath125 is the simplicial complex homeomorphic to @xmath126 depicted in figure [ fig : assoc ] . is the dual of the boundary of the associahedron @xmath127[fig : assoc ] ] the curious reader will discover that the literature on the combinatorics of simplicial complexes is filled with references to the associahedron . in particular , the reader may wish to compare figure [ fig : assoc ] with figure 10 of @xcite , which illustrates the role that the associahedron will play in this paper .
indeed , it is used to describe both the finite rigid subsets sets of @xmath128 and of @xmath129 , illustrating the well - known isomorphism between these two complexes @xcite . in the discussion that follows
we will encounter a number of maps between simplicial complexes which are not simplicial maps but are nonetheless very well - behaved piecewise linear maps .
for example we will describe below a continuous map from the arc complex at infinity @xmath130 to the curve complex @xmath131 which induces a homotopy equivalence ( see known result [ fact : accc ] ) . here
we develop some convenient language and observations for working with these maps .
if @xmath67 is a simplicial complex with vertex set @xmath132 we will specify a general point @xmath133 as a linear combination @xmath134 where @xmath135 and for all @xmath136 we have @xmath137 .
for a point @xmath133 the _ support _ of @xmath138 is the set @xmath139 and for a subset @xmath140 we set @xmath141 .
our simplicial complexes will all be finite dimensional so we will have @xmath142 .
conversely , given a set @xmath143 we will define the _ span _ of @xmath144 to be the set @xmath145 [ def : inter ] let @xmath67 and @xmath146 be simplicial complexes with vertex sets @xmath132 and @xmath147 respectively .
let @xmath148 be a function .
we say that @xmath149 is _ interpolable _ if for any simplex @xmath150 of @xmath67 there is some simplex @xmath151 of @xmath146 such that for every vertex @xmath152 of @xmath150 we have @xmath153 .
note that it suffices to verify the condition of definition [ def : inter ] for the maximal simplices of a finite dimensional simplicial complex @xmath67 .
[ def : lin ] let @xmath67 and @xmath146 be simplicial complexes with vertex sets @xmath132 and @xmath147 respectively .
we say that @xmath154 is a _ linear interpolation _
if the restriction @xmath155 is interpolable and for any point @xmath156 we have @xmath157 note that any interpolable function @xmath148 from the @xmath31-skeleton @xmath132 of a simplicial complex @xmath67 to a simplicial complex @xmath146 extends to a unique linear interpolation @xmath158 via the equation @xmath159 [ lem : lincont ] let @xmath154 be a linear interpolation . then @xmath160 is continuous .
this follows from the gluing lemma from basic topology and the observation that linear interpolations agree on intersections of simplices .
[ lem : quil ] let @xmath161 be linear interpolations and suppose that for each simplex @xmath150 of @xmath67 there is a simplex @xmath151 of @xmath146 such that for each vertex @xmath152 of @xmath150 we have @xmath162
. then @xmath160 and @xmath163 are homotopic functions .
for @xmath164 $ ] define @xmath165 by setting @xmath166 for each vertex @xmath136 .
note that @xmath167 is interpolable so we may extend it to a linear interpolation @xmath168 .
now define @xmath169 by setting @xmath170 .
one observes that @xmath171 and @xmath172 for all @xmath173 .
the continuity of @xmath174 follows from the continuity of the restriction @xmath175 for each simplex @xmath150 of @xmath67 and the gluing lemma .
[ cor : drop ] let @xmath154 be a linear interpolation . for each vertex @xmath136 choose a vertex @xmath176 and define @xmath177 by setting @xmath178 .
then @xmath179 is interpolable and @xmath160 is homotopic to the linear interpolation @xmath180 of @xmath179 .
let @xmath150 be a simplex of @xmath67 .
then since @xmath160 is a linear interpolation there is a simplex @xmath151 of @xmath146 such that @xmath181 and hence @xmath182 . by construction
if @xmath152 is a vertex of @xmath150 then @xmath183 so @xmath184 .
it follows that @xmath179 is interpolable with linear interpolation @xmath185 .
moreover , @xmath186 so we may apply lemma [ lem : quil ] to conclude that @xmath187 and @xmath160 are homotopic . finally we note that if @xmath67 is a simplicial complex then its _ barycentric subdivision _
@xmath188 is the simplicial complex with @xmath31-skeleton given by the set @xmath189 and with @xmath19-simplices given by flags of simplices of @xmath67 of length @xmath190 .
one gets a homeomorphism between @xmath67 and @xmath188 by letting @xmath191 be the function sending the vertex @xmath192 to the point @xmath193 and letting @xmath194 be the linear interpolation of @xmath195 .
the _ barycentric subdivision map _
@xmath196 is the inverse of @xmath197 .
we end our general discussion of linear interpolations with the following observation : [ rem : simp ] a linear interpolation @xmath154 is a simplicial map if and only if for all @xmath136 we have @xmath198 . with the language of linear interpolations we now review harer s homotopy equivalence between the arc complex at infinity and the curve complex .
suppose that @xmath72 and @xmath88 or @xmath54 and @xmath71 .
let @xmath199 and @xmath200 .
let @xmath201 be the barycentric subdivision of @xmath202 .
the vertices of @xmath201 will correspond to non - filling arc systems with @xmath23-simplices corresponding to increasing flags of non - filling arc systems of length @xmath24 . when an arc system @xmath203 does not fill the surface @xmath10 we can associate a curve system @xmath204 to it by letting @xmath205 be a closed regular neighborhood of the union of the arcs in the arc system @xmath203 and letting the curve system @xmath204 be the set of essential ( in @xmath10 ) boundary curves of the subsurface @xmath206 with duplicate curves removed .
let @xmath207 be the function sending the non - filling arc system @xmath203 to the barycenter @xmath208 of the simplex in @xmath68 corresponding to the the curve system @xmath204 .
one sees that @xmath209 is interpolable as follows .
if @xmath150 is a simplex of @xmath201 then it corresponds to an increasing flag of non - filling arc systems @xmath210 .
we may arrange that the corresponding regular neighborhoods satisfy @xmath211 when @xmath212 .
thus all boundary curves can me made to be simultaneously disjoint . let @xmath151 be the simplex of @xmath68 corresponding to the curve system which is the union of all such sets of boundary curves . if @xmath152 is a vertex of @xmath150 then @xmath213
. thus @xmath209 is interpolable .
let @xmath214 be the linear interpolation of @xmath209 .
[ fact : accc ] suppose that @xmath72 and @xmath88 or @xmath54 and @xmath71 . then the map @xmath215 in ( [ eqn : phi ] ) induces a homotopy equivalence between @xmath216 and @xmath131 .
we now restrict to the cases where either @xmath54 and @xmath65 or @xmath69 and @xmath49 and utilize techniques similar to those in @xcite to give a @xmath13-module resolution of @xmath217 .
it is the failure of ( [ eq : tail ] ) below when @xmath72 and @xmath93 which renders our argument invalid for that case . by known result [ fact : con ] the arc complex @xmath218 is contractible and by known result [ fact : accc ]
the arc complex at infinity @xmath219 is homotopy equivalent to the curve complex @xmath22 .
known result [ fact : cc ] says that curve complex has the homotopy type of a wedge of @xmath76-dimensional spheres .
therefore for @xmath220 @xmath221 now consider the cellular chain complex @xmath222 for the pair of spaces @xmath223 .
@xmath224 where as usual we define @xmath225 .
a chain complex is an exact sequence when all of its homology groups are @xmath31 .
note that every homology group of the pair of spaces @xmath223 is zero except for @xmath226 so the chain complex @xmath227 is very close to being an exact sequence .
no arc system with @xmath228 or fewer arcs can fill the surface @xmath229 and no arc system with @xmath230 or fewer arcs can fill the surface @xmath39 . hence in both cases no arc system with @xmath99 or fewer arcs can fill the surface so the entire @xmath76-skeleton of @xmath231 is contained in @xmath202 . thus @xmath232 from equations ( [ eq : sphhom ] ) and ( [ eq : tail ] ) we have that @xmath233 where @xmath234 is the boundary operator for @xmath222 or @xmath235 depending on the context .
we thus have the following very useful description of the homology of the curve complex : [ prop : stpres ] let @xmath236 and @xmath65 or @xmath72 and @xmath49 .
let @xmath76 be as in ( [ eq : tau ] ) and @xmath235 and @xmath237 be the simplicial chain complexes of the arc complex @xmath90 and arc complex at infinity @xmath238 respectively .
then the steinberg module @xmath239 has the following presentation : @xmath240 proposition [ prop : stpres ] says that the reduced homology of the curve complex is the quotient of the free abelian group on all arc systems in @xmath10 with @xmath98 arcs modulo all _ non - filling _ arc systems with @xmath98 arcs and boundaries of arc systems with @xmath241 arcs .
notice that if @xmath242 then by equation ( [ eq : sphhom ] ) we have @xmath243 so @xmath244 is an _ exact _ sequence . as with any terminating exact sequence
we can append two more terms to get a slightly longer exact sequence @xmath245 from ( [ eq : stdes ] ) we know that the penultimate term of this sequence is isomorphic to @xmath217 . since the sequence in ( [ eq : res ] )
is exact we have a @xmath13-module resolution of @xmath217 .
note that a filling arc system in @xmath10 always has a finite but possibly non - trivial stabilizer in @xmath13 .
hence for @xmath246 the relative chain group @xmath247 is not quite a free @xmath13-module .
however , if one restricts to a torsion free subgroup @xmath248 then @xmath247 will be a free @xmath80-module .
[ prop : res ] let @xmath236 and @xmath65 or @xmath72 and @xmath49 .
let @xmath76 be as in ( [ eq : tau ] ) .
the steinberg module @xmath239 has the following @xmath13-module resolution of finite length : @xmath249 this resolution is not free but will be free for any torsion free subgroup @xmath248 .
the terms of this resolution will be finitely generated as @xmath80-modules if the index of @xmath80 in @xmath13 is finite .
we begin the new work in this paper . in this section
we assume that @xmath71 , and let @xmath39 be the surface of genus @xmath31 with marked point set @xmath250 of size @xmath19 .
we turn our attention to the goal of finding a single class in @xmath251 whose orbit under @xmath38 generates @xmath251 as a @xmath252-module . from proposition
[ prop : stpres ] we have that @xmath253 @xmath254 is the free abelian group on all arc systems with @xmath255 arcs .
the mapping class group is not transitive on the set of these arc systems , but there are a finite number of orbits .
we will show that after quotienting by the relations @xmath256 and @xmath257 that every element of @xmath251 will be a linear combination of elements of the @xmath38-orbit of the class of a single arc system .
the set of non - filling arc systems with @xmath255 arcs forms a basis for @xmath256 .
this is a subset of the set of all arc systems with @xmath255 arcs which forms a basis for @xmath254 .
thus we see that non - filling arc systems give the trivial class in @xmath251 .
the union of the arcs in an arc system @xmath203 is a graph @xmath258 .
note that in general the vertices of @xmath258 will be only those marked points in @xmath89 which are endpoints of at least one arc in the arc system @xmath203 .
the arc system @xmath203 fills the surface if @xmath259 is a disjoint union of open disks containing at most one marked point .
if the filling arc system has exactly @xmath255 arcs then each of these disks must contain exactly one marked point .
we note that if @xmath203 is a filling arc system then @xmath258 must be connected since its complement is a disjoint union of open disks .
[ lem : path ] if @xmath203 is a filling arc system in @xmath39 with @xmath255 arcs then its class in the steinberg module @xmath251 is a @xmath252-linear combination of the classes of filling arc systems @xmath260 with @xmath255 arcs such that the graph @xmath261 is a path graph .
let @xmath203 be an arc system in @xmath39 with @xmath255 arcs . if @xmath203 does not fill then its class in @xmath251 is trivial .
tine _ of the graph @xmath258 to be a path subgraph @xmath262 of @xmath258 which is either any single vertex of @xmath258 or else has at least one vertex which has valence one in @xmath258 and exactly two vertices which have valence not equal to 2 in @xmath258 .
( see sketch ( ii ) of figure [ fig : tine ] ) .
note that every vertex of @xmath258 is a tine of length @xmath31 and that distinct tines of @xmath258 can intersect in at most a single vertex .
define the complexity @xmath263 of the arc system @xmath203 to be the length of a longest tine of @xmath258 .
let @xmath264 be a tine of @xmath203 with maximal length .
if @xmath265 then @xmath258 is a path graph and we are done .
assume that @xmath266 .
if @xmath267 let @xmath152 be the unique vertex of @xmath264 . otherwise let @xmath152 be the unique vertex of @xmath264 which has valence 1 in @xmath258 .
the vertex @xmath152 must be in the closure of at least one component of @xmath268 .
let @xmath269 be the marked point contained in the interior of such a component .
( see sketch ( ii ) of figure [ fig : tine ] ) .
let @xmath270 be an arc disjoint from the arcs of @xmath203 connecting @xmath152 to @xmath269 .
let @xmath271 which is an arc system with @xmath272 arcs .
( see sketch ( iii ) of figure [ fig : tine ] ) .
given @xmath273 set @xmath274 .
we claim exactly one of the following holds : 1 .
2 . @xmath276 is a non - filling arc system in @xmath39 .
3 . @xmath276 is a filling arc system with @xmath277 .
note that @xmath275 if and only if @xmath278 .
suppose @xmath279 .
then @xmath280 .
if @xmath281 then @xmath282 is disconnected so @xmath276 is non - filling . if @xmath283 then @xmath276 may or may not be a filling system but in either case @xmath284 .
we may use the relation @xmath285 in @xmath251 to express the class of the arc system @xmath203 as a @xmath252-linear combination of classes of arc systems with higher complexity . by induction
we may write any such class as a @xmath252-linear combination of classes of filling arcs systems with the maximal complexity @xmath255 .
the graph for each of these arc systems must be a path graph .
[ prop : gen0 ] suppose @xmath71 .
let @xmath286 be the @xmath287-simplex corresponding to the arc system @xmath288 in figure [ fig : path ] .
then the class @xmath289 \in \operatorname{st}(\surface_0^n ) = \frac{\operatorname{c}_{n-3}(\ac)}{\operatorname{c}_{n-3}(\ac_\infty ) \oplus \partial \operatorname{c}_{n-2}(\ac)}\ ] ] is non - trivial and generates @xmath251 as a @xmath38-module . with @xmath255 arcs in the surface @xmath39 whose class generates @xmath251 as a @xmath38-module.[fig : path],title="fig : " ] ( -209,9)@xmath290 ( -173,9)@xmath291 ( -69,9)@xmath292 ( -9,11)@xmath293 ( -126,-1)@xmath123 by lemma [ lem : path ]
the class of any arc system with @xmath255 arcs is a @xmath252-linear combination of classes of arc systems whose graphs are path graphs . all path graphs with @xmath255 edges in @xmath39 are in the @xmath38-orbit of the arc system @xmath288 pictured in figure [ fig : path ] .
hence the @xmath38-orbit of the class of @xmath294 generates @xmath251 as a @xmath252-module . in other words
@xmath295 $ ] generates @xmath251 as a @xmath38-module .
suppose that @xmath295 $ ] is the trivial class in @xmath251 .
then @xmath251 is the zero module .
let @xmath296 be a torsion free finite index subgroup .
then by ( * ? ? ?
* theorem 4.1 ) the cohomological dimension of @xmath80 is @xmath230 .
we may use duality to show that for all @xmath23 and all @xmath80-modules @xmath83 @xmath297 this implies that the cohomological dimension of @xmath80 is @xmath31 .
hence we have a contradiction .
it follows that @xmath295 $ ] is a non - trivial class in @xmath251 .
again we assume that @xmath54 and @xmath65 .
in proposition [ prop : gen0 ] we have a non - trivial class in @xmath298 we will now use the homotopy equivalence between @xmath202 and @xmath68 and the techniques of @xmath34[sec : simpmap ] to get a @xmath36-sphere in @xmath68 representing this non - trivial class . after simplification our @xmath36-sphere will be the dual @xmath299 of the boundary of the associahedron @xmath300 ( see theorem [ thm : assoc ] ) . the class of the filling arc system in figure [ fig : path ] is represented by a single @xmath287-simplex @xmath301 all of whose proper faces are contained in the arc complex at infinity .
the connecting homomorphism @xmath302 sends the class @xmath295 \in \operatorname{h}_{n-3}(\ac , \ac_\infty;\z)$ ] to the class @xmath303 \in \widetilde{\operatorname{h}}_{n-4}(\ac_\infty;\z)$ ] where @xmath304 denotes the @xmath36-skeleton of the @xmath287-simplex @xmath305 .
let @xmath306 be the barycentric subdivision map from ( [ eq : bs ] ) . recall that the homotopy equivalence @xmath307 from ( [ eqn : phi ] ) sends a non - filling arc system to the collection of boundary curves of a regular neighborhood of the arc system .
it follows that the homology class of the @xmath36-sphere @xmath308 is a @xmath38-module generator for @xmath309 and is non - trivial .
there are a number of properties of the @xmath36-sphere @xmath310 which are undesirable .
firstly , since @xmath215 is defined on the barycentric subdivision of @xmath202 the sphere @xmath308 has @xmath311 vertices which is exponential in the number of marked points .
secondly , it is a subcomplex of the barycentric subdivision of the curve complex , whereas one would like to describe the sphere as a subcomplex of the curve complex itself . in order to address these issues
we will replace the sphere @xmath310 with a simpler one representing the same free homotopy class .
in fact our simplified sphere @xmath312 described below will agree with @xmath310 setwise but will have a much simpler structure as a simplicial complex .
let @xmath313 be the flag complex whose vertices correspond to non - empty proper sub arc systems of the filling arc system @xmath288 in figure [ fig : path ] and whose @xmath23-simplices correspond to increasing flags of such sub arc systems of length @xmath24 .
notice that @xmath314 is the homeomorphic image of @xmath315 under the barycentric subdivision map @xmath306 from ( [ eq : bs ] ) and hence topologically a sphere of dimension @xmath316 .
let @xmath317 be the inclusion map and let @xmath318 where @xmath319 is as in ( [ eqn : phi ] ) .
now we will apply the techniques of @xmath34[sec : simpmap ] to simplify @xmath320 .
let @xmath147 be the @xmath31-skeleton of @xmath146 .
define the function @xmath321 as follows .
let @xmath322 be a vertex of @xmath146 .
set @xmath323 to be the left - most connected component of the arc system @xmath203 and @xmath324 to be the unique boundary curve of @xmath325 .
notice that @xmath326 .
hence by corollary [ cor : drop ] the function @xmath327 is interpolable .
let @xmath328 be its linear interpolation which by corollary [ cor : drop ] is homotopic to @xmath320 .
define the subcomplex @xmath329 as follows . for
each non - empty interval @xmath330 let @xmath331 be the boundary curve of the arc system @xmath332 .
let @xmath333 let @xmath334 .
the reader may now recognize the subcomplex @xmath67 as precisely the complex identified by aramayona and leininger in ( * ? ? ?
* @xmath343 ) . for completeness of exposition
we include lemma [ lem : xassoc ] below and in @xmath34[sec : al ] below rectify the description of @xmath67 above with that given in ( * ? ? ?
* @xmath343 ) .
[ lem : xassoc ] the complex @xmath67 and the simplicial complex @xmath299 ( see theorem [ thm : assoc ] ) are isomorphic as simplicial complexes .
let @xmath335 be the graph in figure [ fig : pathonly ] .
if @xmath336 are two non - empty intervals then the boundary curves @xmath337 and @xmath338 form a curve system exactly when @xmath339 and @xmath340 are compatible tubes for the graph @xmath335 .
we caution the reader not to confuse the graph @xmath335 which has a _ vertex _ for each arc in the arc system @xmath288 and the underlying graph @xmath341 for the arc system @xmath288 from @xmath34[sec : surfaces of genus 0 ] in which each arc of @xmath288 is an _
edge_. the linear interpolation @xmath342 of the map sending the tube @xmath343 of @xmath335 to the curve @xmath344 is a simplicial map by remark [ rem : simp ] and its inverse is the linear interpolation of the function sending the curve @xmath331 to the maximal subinterval @xmath345 such that the arc system @xmath346 is entirely contained in the component of @xmath347 which does not contain the marked point @xmath348 .
[ lem : ytox ] the map @xmath349 is a simplicial map , @xmath350 and @xmath351 is a homotopy equivalence .
the map @xmath349 is a simplicial map onto its image by remark [ rem : simp ] and the observation that for each vertex @xmath203 of @xmath146 we have @xmath352 if @xmath203 is a vertex of @xmath146 then @xmath353 is a vertex of @xmath67 .
the map @xmath349 is simplicial
so @xmath354 .
observe that @xmath146 is topologically an @xmath36-sphere and by lemma [ lem : xassoc ] that @xmath67 is topologically an @xmath36-sphere .
hence the map @xmath355 will be a homotopy equivalence if and only if it is a degree @xmath356 map .
let @xmath357 then @xmath358 where @xmath359 .
thus the degree of @xmath349 is equal to the degree of the induced map @xmath360 where @xmath361 denotes the link of the vertex @xmath269 in the complex @xmath362 .
the map @xmath363 is exactly the map @xmath349 for the surface with one less puncture . for the base case surface
@xmath25 the map @xmath364 is a homeomorphism of @xmath31-spheres which has degree @xmath356 .
hence , inductively , for all @xmath71 the map @xmath364 has degree @xmath356 and is therefore a homotopy equivalence .
it follows that @xmath355 is surjective .
we have arrived at the following simplification of our @xmath38-module generator for @xmath309 .
[ prop : simpsphere ] assume @xmath71 .
let @xmath329 be the subcomplex of all simplices of @xmath68 whose vertices are boundary curves of connected sub arc systems of the arc system @xmath288 in figure [ fig : path ] . then as a simplicial complex @xmath67 is the dual @xmath299 of the boundary of the associahedron @xmath300 ( see theorem [ thm : assoc ] ) and with a proper choice of orientation the class @xmath365 \in \widetilde{\operatorname{h}}_{n-4}(\cc;\z)$ ] is the same as that of @xmath294 from proposition [ prop : gen0 ] and hence is a non - trivial generator for @xmath309 as a @xmath38-module .
lemma [ lem : xassoc ] shows that @xmath67 and @xmath299 are isomorphic simplicial complexes .
the map @xmath366 above represents a @xmath38-module generator for @xmath309 . as we noted @xmath367 is homotopic to @xmath320 and thus represents the same class .
let @xmath368 be the inclusion map .
by lemma [ lem : ytox ] the map @xmath351 has a homotopy inverse @xmath180 .
thus @xmath369 is homotopic to @xmath370 which represents the same class as @xmath320 in @xmath309 . in @xcite aramayona and leininger
show that certain finite subcomplexes @xmath371 of the curve complex are _ rigid sets_. that is , any injective simplicial map @xmath372 extends to a simplicial injection @xmath373 and hence is induced by a mapping class . in the case that @xmath374
their finite rigid set @xmath375 is a topological sphere of dimension @xmath316 and they ask ( * ? ? ?
* question 2 ) if the class @xmath376 \in \widetilde{\operatorname{h}}_{n-4}(\cc ( \surface_0^n))$ ] is non - trivial .
the construction of the set @xmath377 is as follows : let @xmath378 be a regular @xmath19-gon and let @xmath39 be the double of @xmath378 along its boundary with each vertex of @xmath378 giving a marked point
. for each line segment @xmath269 joining two nonadjacent sides of @xmath378 the double @xmath379 of @xmath269 will be an essential curve in @xmath39 .
let @xmath380 be the set of all @xmath381 essential curves in @xmath39 which may be constructed in this manner .
define @xmath382 to be the union of all simplices of @xmath383 whose vertices all lie in the set @xmath380 .
the reader is referred to @xmath34[sec : intro ] of this paper for the statement of theorem [ thm : same ] . using proposition [ prop : simpsphere ]
, we are now ready to give the proof .
assume that @xmath55 .
arrange the arc system in figure [ fig : path ] so that it is entirely contained in the image of the boundary of the @xmath19-gon @xmath378 in @xmath39 .
arrange the arcs in figure [ fig : path ] so that they coincide with all but two adjacent edges of @xmath378 .
let @xmath348 be the marked point of @xmath39 which is not incident with any arcs .
if @xmath269 is a line segment joining two nonadjacent sides of @xmath378 then its double @xmath379 will be the boundary curve for the maximal connected sub arc system of the arc system in figure [ fig : path ] which is disjoint from @xmath379 and which is in the component of @xmath384 which does not contain the marked point @xmath348 .
this pairing identifies the vertices of @xmath385 and @xmath67 . since both subcomplexes are the full subcomplexes of @xmath386 spanned by their vertices the subcomplexes must coincide .
we address the case of the surface @xmath25 in @xmath34[ss : sporadic ] below .
we now turn our attention to the surface @xmath10 where @xmath72 and @xmath389 .
our main interest is in the case @xmath390 , but in the process of investigating that we will need to study the case @xmath49 .
to get started , we review what was done in @xcite to identify the homologically non - trivial sphere in @xmath75 described there . after that
, we use the techniques of @xmath34[sec : simpmap ] to radically simplify this sphere . finally in @xmath34[sec :
gcomp ] we compare this simplified sphere to the finite rigid sets for @xmath10 given in @xcite .
consider the surface of genus @xmath163 with @xmath120 marked point depicted in figure [ fig : salgen ] .
let @xmath392 be the @xmath393-simplex corresponding to the filling arc system @xmath394 given in figure [ fig : salgen ] .
in @xcite it is shown that the class @xmath395 $ ] generates @xmath396 as a @xmath397-module . with @xmath398 arcs in the surface @xmath229
whose class generates @xmath399 as a @xmath397-module.[fig : salgen],title="fig : " ] ( -64,97 ) ( -8,97 ) ( -219,72)@xmath400 ( -215,135)@xmath401 ( -170,118)@xmath402 ( -187,160)@xmath403 ( -145,156)@xmath404 ( -48,156)@xmath405 ( -78,142)@xmath406 ( -92,60)@xmath407 ( -80,30)@xmath408 ( -173,57)@xmath409 ( -190,34)@xmath410 now consider the surface @xmath411 without marked points .
here we have no arc complex , but by a result of harer ( * ? ? ?
* lemma 3.6 ) following from more general work by kent , leininger and schleimer ( * ? ? ?
* corollary 1.1 ) the map @xmath412 which `` forgets '' the marked point is a homotopy equivalence .
hence we may identify @xmath399 and @xmath413 via the induced map @xmath414 for all @xmath72 . to continue , suppose that @xmath72 and @xmath49 .
let @xmath415 be the flag complex whose vertices correspond to non - empty proper sub arc systems of the filling arc system @xmath394 in figure [ fig : salgen ] and whose @xmath23-simplices correspond to increasing flags of such sub arc systems of length @xmath24 .
the simplicial complex @xmath416 is the barycentric subdivision of the boundary of a @xmath393-simplex and hence topologically a sphere of dimension @xmath417 .
again we let @xmath418 be the inclusion map and set @xmath318 where @xmath319 is as in ( [ eqn : phi ] ) .
the _ sequential components _ of a sub arc system @xmath203 of the arc system @xmath394 in figure [ fig : salgen ] will be the maximal subsets of @xmath203 of the form @xmath419 where @xmath420 . note that @xmath203 is the disjoint union of its sequential components .
if @xmath421 are sequential components of @xmath203 then we say that @xmath422 _ precedes _
@xmath423 if for all @xmath424 and all @xmath425 we have @xmath426 . under this order we will refer to the _ first
_ , _ second _ , _ last _ , _ etc
. _ sequential component of @xmath203 .
the order on the sequential components of a sub arc system @xmath427 should not be confused with the containment partial order on all arc systems under which the sequential components of @xmath203 are incomparable .
a sequential component will be called _ odd _ if it has an odd number of elements and _ even _ if it has an even number of elements .
the first step in the simplification of @xmath428 will be to define a map @xmath429 which will prove to be homotopic to @xmath320 .
( see lemma [ lem : hom ] below ) .
let @xmath430 be the @xmath31-skeleton of @xmath416 .
define the function @xmath431 as follows .
let @xmath432 be a proper sub arc system of @xmath394 and let @xmath433 be the first sequential component of @xmath203 .
let @xmath434 be a closed regular neighborhood of @xmath435 in @xmath229 . if @xmath436 is an odd sequential component set @xmath437 and let @xmath324 be the boundary component of @xmath438 which is to the right when following @xmath439 in the clockwise direction .
( see sketch ( i ) of figure [ fig : oesc ] ) .
if @xmath433 is an even sequential component let @xmath440 where @xmath82 is a closed regular neighborhood of the marked point in @xmath441 whose intersection with the boundary curve of @xmath434 is a single line segment and which contains @xmath442 for each @xmath443 satisfying @xmath444 or @xmath445 .
let @xmath324 be the unique boundary component of @xmath438 .
( see sketch ( ii ) of figure [ fig : oesc ] ) .
let @xmath446 be two proper sub arc systems of the arc system @xmath394 in figure [ fig : salgen ] with respective first sequential components @xmath436 and @xmath448 .
then either @xmath449 or @xmath450 has two sequential components with @xmath448 preceding @xmath436 .
suppose @xmath449 .
if @xmath448 is an odd sequential component or both @xmath436 and @xmath448 are even sequential components then we may choose the closed regions @xmath438 and @xmath451 so that @xmath452 .
hence the curves @xmath324 and @xmath453 are disjoint and form a curve system .
again suppose @xmath449 .
if @xmath448 is an even sequential component and @xmath436 is an odd sequential component then we may choose the closed region @xmath438 and a closed regular neighborhood @xmath454 of @xmath455 so that @xmath456 .
we may choose the closed disk neighborhood @xmath82 of the marked point so that it is disjoint from the boundary curve @xmath324 of @xmath438 and set @xmath457 .
we again see that the curves @xmath324 and @xmath453 are disjoint and form a curve system .
suppose @xmath450 has two sequential components .
if at least one of @xmath436 and @xmath448 is an even sequential component then we may arrange for the regions @xmath438 and @xmath451 to be disjoint .
thus the curves @xmath324 and @xmath453 are disjoint and form a curve system .
if both @xmath436 and @xmath448 are odd sequential components then we may choose @xmath438 and @xmath451 so that their intersection is a bigon containing the marked point which is disjoint from @xmath324 and @xmath453 .
again we see that the curves @xmath324 and @xmath453 are disjoint .
recall that a set of curves forms a curve system if and only if every subset of size two is a curve system .
let @xmath150 be a simplex of @xmath416 with vertex set @xmath458 .
then every subset of @xmath459 of size two is of the form @xmath447 for some arc systems @xmath460 with @xmath461 .
by lemma [ lem : self ] all such sets of size two are curve systems .
it follows that @xmath459 is a curve system and that @xmath327 is interpolable .
let @xmath462 be its linear interpolation .
let @xmath446 be two proper sub arc systems of the arc system @xmath394 .
then we may choose a regular neighborhood @xmath466 of @xmath467 in @xmath229 and a closed region @xmath438 so that @xmath468 .
the curves in @xmath469 may be chosen to be the pairwise disjoint boundary curves of @xmath466 and the single curve in @xmath470 may be taken to be the boundary curve of @xmath438 .
therefore the curves in @xmath471 are pairwise disjoint and form a curve system .
let @xmath150 be a simplex of @xmath416 with vertex set @xmath472 satisfying @xmath473 .
consider the simplicial complex @xmath474 which is a prism triangulated by @xmath24 simplices of dimension @xmath24 with vertex sets : @xmath475 the @xmath31-skeleton of @xmath262 is the set @xmath476 .
let @xmath477 be the function @xmath478 we wish to show that @xmath479 is interpolable .
each maximal simplex of @xmath474 has vertex set @xmath480 for some @xmath443 .
thus we must show that @xmath481 is a curve system for each @xmath443 . recall that a set of curves is a curve system if and only if each subset of size two is a curve system .
a subset of size two of the set @xmath481 must be a subset of a set of the form @xmath482 , a set of the form @xmath483 , or a set of the form @xmath484 for some curve systems @xmath485 .
the set @xmath486 is a curve system by lemma [ lem : self ] .
the set @xmath487 is a curve system by lemma [ lem : lower ] .
finally the set @xmath488 is a curve system since it consists of boundary curves of two closed subsurfaces @xmath489 and @xmath466 of @xmath229 with @xmath490 .
hence @xmath479 is interpolable .
let @xmath491 be its linear interpolation .
by the gluing lemma these @xmath492 s assemble to a homotopy @xmath493 from @xmath349 to @xmath320 .
for each interval @xmath502 let @xmath503 be a closed regular neighborhood of the set @xmath504 .
if @xmath505 is even then @xmath503 has one boundary component .
let @xmath331 be that boundary component .
if @xmath505 is odd then @xmath503 has two boundary components . in this case
let @xmath331 be the boundary component of @xmath503 in the `` back half '' ( if @xmath197 is even ) or `` top half '' ( if @xmath197 is odd ) of the surface .
let @xmath132 be the set @xmath506 and define @xmath507 to be the subcomplex of all simplices of @xmath508 whose vertices are in @xmath132 .
let @xmath509 be the image of @xmath510 under the map @xmath511 from ( [ eq : psi ] ) above which `` forgets '' the marked point .
for intervals @xmath514 the curves @xmath515 and @xmath516 are disjoint precisely when the tubes @xmath517 and @xmath518 for the graph @xmath519 in figure [ fig : pathonly ] are compatible .
let @xmath520 be the @xmath31-skeleton of the dual @xmath513 of the boundary of the associahedron @xmath521 . the bijection @xmath522 given by @xmath523 and its inverse are interpolable maps showing that @xmath510 and @xmath513 are isomorphic as simplicial complexes .
the map @xmath524 is an isomorphism of simplicial complexes .
[ lem : ztox ] assume @xmath72 and @xmath47 .
the maps @xmath464 and @xmath525 are simplicial maps .
the images of @xmath349 and @xmath526 are @xmath510 and @xmath527 respectively . furthermore @xmath528 and @xmath529 are homotopy equivalences .
the map @xmath464 is a simplicial map onto its image by remark [ rem : simp ] and the observation that for each vertex @xmath203 of @xmath416 we have @xmath530 the map @xmath531 a simplicial map so the composition @xmath526 is simplicial .
the curves in the set @xmath132 given in ( [ eq : xdef ] ) are precisely the curves in the image of @xmath349 .
thus if @xmath203 is a vertex of @xmath416 then @xmath353 is a vertex of @xmath510 and @xmath532 is a vertex of @xmath527 .
the maps @xmath349 and @xmath526 are simplicial so @xmath533 and @xmath534 .
observe that @xmath416 is topologically a @xmath50-sphere and by lemma [ lem : gassoc ] that @xmath510 is topologically an @xmath50-sphere .
moreover , as maps between abstract simplicial complexes the map @xmath535 agrees with the map @xmath536 from lemma [ lem : ytox ] .
hence the map @xmath537 is a homotopy equivalence and has degree @xmath356 .
it follows that @xmath537 is surjective .
the restriction @xmath524 is an isomorphism of simplicial complexes and hence a degree @xmath356 map .
it follows that @xmath529 has degree @xmath356 .
it is therefore a surjective simplicial homotopy equivalence .
[ prop : geng ] let @xmath72 and @xmath538 .
let @xmath132 be the set of essential curves in @xmath10 given in ( [ eq : xdef ] ) and @xmath539 be the full subcomplex of @xmath75 generated by @xmath132 .
then as a simplicial complex @xmath512 is isomorphic to the dual @xmath513 of the boundary of the associahedron @xmath521 ( see theorem [ thm : assoc ] ) .
furthermore @xmath512 is topologically a sphere of dimension @xmath417 and , choosing an orientation , the class @xmath540 \in \widetilde{\operatorname{h}}_{2g-2}(\cc(\surface_g^n);\z)$ ] is non - trivial and generates @xmath541 as a @xmath13-module .
lemma [ lem : gassoc ] shows that @xmath510 , @xmath527 and @xmath513 are isomorphic simplicial complexes . the map @xmath463 ( respectively @xmath542 ) above represents a @xmath397-module ( respectively @xmath543-module ) generator for @xmath544 ( respectively @xmath545 ) . as we noted @xmath349 is homotopic to @xmath320 so @xmath526 is homotopic to @xmath546 .
let @xmath547 and @xmath548 be the inclusion maps . by lemma [ lem : ztox ] the maps @xmath528 and @xmath529 have homotopy inverses @xmath549 and @xmath550 respectively .
thus @xmath551 and @xmath552 are homotopic to @xmath553 and @xmath554 respectively which represent the same classes as @xmath555 in @xmath556 and @xmath546 in @xmath557 respectively .
again let @xmath72 and @xmath47 . in @xcite a finite rigid set @xmath558 is given .
our essential sphere @xmath512 is a subset of @xmath559 .
given the surprising coincidence of our essential sphere and the finite rigid set of aramayona and leininger for the surface @xmath39 one might be tempted to conjecture that a subcomplex of the curve complex is essential if and only if it is finitely rigid . as suggested by theorem [ thm : notiff ]
this is not the case .
see @xmath34[sec : intro ] of this paper for the statement of theorem [ thm : notiff ] .
suppose that @xmath46 and @xmath47 .
let @xmath132 be the set of curves described in ( [ eq : xdef ] ) .
let @xmath560 be a closed regular neighborhood of @xmath561 where the curves @xmath562 are as in figure [ fig : chain ] above .
then @xmath560 has two boundary components .
one boundary component @xmath563 of @xmath560 is in @xmath132 the other @xmath564 is not .
let @xmath565 be the function @xmath566 every curve in @xmath132 that is disjoint from @xmath563 is also disjoint from @xmath564 .
thus @xmath149 is interpolable with linear interpolation @xmath567 .
the map @xmath160 is simplicial since @xmath568 for all @xmath569 .
the map @xmath160 is an injection since if @xmath570 are distinct disjoint curves then @xmath571 and @xmath572 are distinct disjoint curves .
the map @xmath160 can not be induced by an extended mapping class since @xmath563 has non - empty intersection with @xmath573 but @xmath574 and @xmath575 are disjoint . if @xmath48 and @xmath49 then @xmath576 .
again , we claim that @xmath67 is not a finite rigid set .
let @xmath132 be the set of curves described in ( [ eq : xdef ] ) .
let @xmath577 be a closed regular neighborhood of @xmath578 where the curves @xmath497 and @xmath579 are as in figure [ fig : chain ] above .
then @xmath577 has a single boundary component @xmath580 .
fix a simple path @xmath138 from @xmath580 to the marked point which is disjoint from all other curves in @xmath132 .
let @xmath581 be the boundary component of a regular neighborhood of @xmath582 which is not isotopic to @xmath580 .
let @xmath565 be the function @xmath583 the function @xmath149 preserves disjointness of curves .
thus @xmath149 is interpolable with continuous linear interpolation @xmath567 .
the map @xmath160 is simplicial since @xmath568 for all @xmath569 .
the map @xmath160 is an injection since if @xmath570 are distinct disjoint curves then @xmath571 and @xmath572 are distinct disjoint curves . we claim that @xmath160 is not induced by an extended mapping class @xmath584 .
note that @xmath160 is the identity on the chain curves @xmath585 .
suppose that the mapping class @xmath584 extends @xmath160 .
then without loss of generality we may represent the class of @xmath187 by a map which induces an isometry on the regular @xmath586-gon @xmath587 one gets by cutting along all of the chain curves .
there are 4 possibilities for @xmath187 generated by the orientation reversing reflection about the plane through the even curves @xmath588 and the orientation reversing reflection of the surface about the plane through the odd curves @xmath589 . of these @xmath590 maps
only the identity map fixes both @xmath563 and @xmath591 .
since @xmath160 fixes both @xmath563 and @xmath591 but is not the identity it can not be induced by an extended mapping class .
we address the `` missing cases '' in theorems [ thm : same ] and [ thm : notiff ] .
the case of @xmath592 was included in the statement of theorem 2 , but @xmath593 was omitted . indeed , as noted by aramayona and leininger (
* @xmath343.1 ) , we have @xmath594 and hence @xmath595 is a finite rigid set . recall that from the group cohomological point of view our nonstandard definition of the curve complexes @xmath28 , @xmath29 and @xmath30 as countably infinite discrete sets is appropriate . in particular , propositions [ prop : simpsphere ] or [ prop : geng ] hold for these cases .
clearly these complexes have no finite rigid subsets .
in particular our @xmath31-spheres @xmath596 , @xmath597 and @xmath598
can not be finite rigid subsets of their respective curve complexes . however , in these three sporadic cases if @xmath599 is the standard 2-dimensional curve complex with @xmath31-skeleton given by the set of essential simple closed curves and a @xmath23-simplex for each set of @xmath24 curves with minimal pairwise intersection number , then our homologically non - trivial spheres in @xmath68 can be associated with rigid sets in @xmath599 .
let @xmath600 be the subcomplex of @xmath599 spanned by @xmath67 .
the finite rigid set @xmath385 of aramayona and leininger ( * ? ? ?
* @xmath344 ) is a maximal 2-simplex and a simplicial injection of @xmath385 into @xmath599 is induced by a unique extended mapping class .
we may take our finite rigid set @xmath600 to be a proper subcomplex of @xmath385 .
a simplicial injection of @xmath600 into @xmath599 is induced by a unique mapping class . in this case
@xmath600 is not a sphere , but a cell of dimension one .
j. aramayona and c. leininger , finite rigid sets in curve complexes , j. topol . anal . * 5 * ( 2013 ) , no . 2 , 183203 .
n. broaddus , homology of the curve complex and the steinberg module of the mapping class group , duke math .
j. * 161 * ( 2012 ) , no . 10 , 19431969 .
mr2954621 k. s. brown , _ cohomology of groups _ ,
springer , new york , 1982 .
w. j. harvey , boundary structure of the modular group , in _ riemann surfaces and related topics :
proceedings of the 1978 stony brook conference ( state univ .
new york , stony brook , n.y . , 1978 ) _
, 245251 , ann . of math .
, 97 , princeton univ . press ,
princeton , n.j .
a. hatcher , on triangulations of surfaces , topology appl .
* 40 * ( 1991 ) , no . 2 , 189194 . updated version available at ` http://www.math.cornell.edu/~hatcher/papers/triangsurf.pdf ` n. v. ivanov , automorphism of complexes of curves and of teichmller spaces , internat .
res . notices * 1997 * , no . 14 , 651666 . | aramayona and leininger have provided a `` finite rigid subset '' @xmath0 of the curve complex @xmath1 of a surface @xmath2 , characterized by the fact that any simplicial injection @xmath3 is induced by a unique element of the mapping class group @xmath4 . in this paper
we prove that , in the case of the sphere with @xmath5 marked points , the reduced homology class of the finite rigid set of aramayona and leininger is a @xmath4-module generator for the reduced homology of the curve complex @xmath1 , answering in the affirmative a question posed in @xcite . for the surface
@xmath6 with @xmath7 and @xmath8 we find that the finite rigid set @xmath0 of aramayona and leininger contains a proper subcomplex @xmath9 whose reduced homology class is a @xmath4-module generator for the reduced homology of @xmath1 but which is not itself rigid . |
concept of compressed sensing ( cs ) was initiated by donoho @xcite , cand@xmath22s , romberg and tao @xcite and cand@xmath22s and tao @xcite with the involved essential idea recovering some original @xmath23-dimensional but sparse signal@xmath24image from linear measurement with dimension far fewer than @xmath23 .
large numbers of researchers , including applied mathematicians , computer scientists and engineers , have paid their attention to this area owing to its wide applications in signal processing , communications , astronomy , biology , medicine , seismology and so on , see , e.g. , survey papers @xcite and a monograph @xcite . to recover a sparse solution @xmath25 of the underdetermined system of the form @xmath26 , where @xmath27 is the available
measurement and @xmath28 is a known measurement matrix ( with @xmath29 ) , the underlying model is the following @xmath30 _ minimization _ : @xmath31 where @xmath32 is @xmath30-norm of the vector @xmath33 , i.e. , the number of nonzero entries in @xmath34 ( this is not a true norm , as @xmath35 is not positive homogeneous ) .
however ( [ l0 ] ) is combinatorial and computationally intractable .
one natural approach is to solve ( [ l0 ] ) via convex _
@xmath4 minimization _ : @xmath36 the other way is to relax ( [ l0 ] ) through the nonconvex _
@xmath37 minimization _ : @xmath38 where @xmath39 . motivated by the fact @xmath40
, it is shown that there are several advantages of using this approach to recover the sparse signal @xcite .
this model for recovering the sparse solution is widely considered , see @xcite .
one of the most popular conditions for exact sparse recovery via @xmath4 or @xmath5 minimization is related to the _ restricted isometry property _ ( rip ) introduced by cand@xmath22s and tao @xcite , which was recalled as follows . for @xmath41 , the restricted isometry
constant is the smallest positive number @xmath42 such that @xmath43 holds for all @xmath3-sparse vector @xmath33 , i.e. , @xmath44 .
it is known that @xmath15 has the monotone property for @xmath3 ( see , e.g. , @xcite ) , i.e. , @xmath45 current upper bounds on the restricted isometry constants ( rics ) via @xmath37 minimization for exact signal recovery were emerged in many studies @xcite , such as @xmath46 for any @xmath47 $ ] in @xcite , @xmath46 for any @xmath48 $ ] with some @xmath49 $ ] in @xcite and @xmath50 for any @xmath51 $ ] in @xcite . comparing with those ric bounds , cai and zhang @xcite
recently have given a sharp bound @xmath52 via @xmath4 minimization .
motivated by results above , we make our concentrations on improving ric bounds via @xmath5 relaxation with @xmath16 .
the main contributions of this paper are the following three aspects : \(i ) if the restricted isometry constant of @xmath53 satisfies @xmath54 for @xmath8 , which implies @xmath52 , then exact recovery can be succeeded via @xmath55 and @xmath7 minimizations .
\(ii ) for any @xmath14 , the bound for @xmath56 is an nondecreasing function on @xmath9 and @xmath12 . moreover , several sufficient conditions are derived , such as for any @xmath9 , @xmath10 when @xmath11 , for any @xmath12 , @xmath13 when @xmath14 . the detailed can be seen in tab . 2 of the section iii , which are all better bounds than current ones in terms of @xmath57 minimization .
\(iii ) the bound on @xmath15 is given as well for any @xmath16 .
especially for @xmath17 , we obtain @xmath18 when @xmath3 is even or @xmath20 when @xmath21 is odd .
the organization of this paper is as follows . in the next section
, we establish several key lemmas .
our main results on @xmath58 with @xmath8 and @xmath15 will be presented in sections iii and iv respectively .
we make some concluding remarks in section v and give the proofs of all lemmas and theorems in the last section .
this section will propose several technical lemmas , which play an important role in the sequel analysis .
we begin with recalling the lemma of the sparse representation of a polytope stated by cai and zhang @xcite . here
, we define @xmath59 and @xmath60 ( in fact , @xmath61 is not a norm since the triangle inequality fails ) .
[ lemma1 ] for a positive number @xmath62 and a positive integer @xmath63 , define the polytope @xmath64 by @xmath65 for any @xmath66 , define the set @xmath67 of sparse vectors by @xmath68 then @xmath69 if and only if @xmath70 is in the convex hull of @xmath71 . in particular , any @xmath69 can be expressed as @xmath72 where @xmath73 next we establish an interesting and important inequality in the following lemma , which gives a sharpened estimation of @xmath4 with @xmath74 and @xmath75 . [ lemma2]for @xmath76 $ ] and @xmath33 , we have @xmath77 where @xmath78 moreover , @xmath79 is a nonincreasing and convex function of @xmath80 $ ] with @xmath81 captypefigure actually , we can substitute @xmath23 with @xmath32 in inequality ( [ l1q ] ) , which leads to @xmath82 moreover , combining with the h@xmath83lder inequality and @xmath84 , we have for @xmath76 $ ] and @xmath33 , we have @xmath85 here , @xmath86 is an interesting inequality .
although @xmath86 will not be applied in our proof , it manifests the relationship between @xmath4 and @xmath5 norm . in order to analyze a sequent useful function more clearly
, we first observe the function @xmath87 of @xmath88 , whose figure is plotted below .
captypefigure it is easy to check that @xmath89 so @xmath87 can be defined as a function of @xmath90 on @xmath91 $ ] , and it is a nondecreasing function . in addition , for any given integer @xmath14 , it is trivial that if @xmath87 is an integer , then @xmath92 apparently is an integer as well for instance @xmath93 .
however , the integrity of @xmath87 is not necessary to ensure the integrity of @xmath92 , such as @xmath94 and @xmath95 .
based on analysis above , we now define a real valued function @xmath96 by @xmath97 where @xmath79 is defined as in ( [ pq ] ) and @xmath98 denotes the smallest integer that is no less than @xmath99 .
[ lemma3 ] let @xmath100 be defined as in @xmath101 .
then @xmath102 when @xmath92 is an integer and otherwise @xmath103 .
moreover , @xmath104 therefore , @xmath100 can be regarded as a function of @xmath90 on @xmath91 $ ] , and the image of @xmath100 with the special case @xmath105 , where @xmath106 , is plotted in fig.[fig3 ] .
captypefigure another two useful functions are introduced and analyzed in the following lemma , which will ease sequent analysis of our main results .
[ lemma4]for @xmath8 and @xmath107 , we define @xmath108 then @xmath109 is a nonincreasing function on @xmath110 when @xmath111 is fixed while a nondecreasing function on @xmath111 when @xmath110 is fixed .
now we give our main results on @xmath58 with @xmath8 : [ theorem]for any @xmath76 $ ] , if @xmath112 holds for some @xmath8 , then each @xmath3-sparse minimizer of the @xmath5 minimization @xmath113 is the sparse solution of @xmath114 .
furthermore , setting @xmath115 with @xmath116 , then the sufficient condition @xmath117 of exact signal recovery can be reformulated as @xmath118 from lemma [ lemma3 ] , when @xmath119 or @xmath92 is an integer ( such as @xmath120 ) , it follows that @xmath102 .
associating with ( [ dertagk ] ) in theorem [ theorem ] , we have @xmath121 .
therefore , a corollary can be elicited as below .
[ prop1 ] for @xmath119 or @xmath88 such that @xmath92 is an integer , if @xmath122 holds with some @xmath8 and @xmath14 , then each @xmath3-sparse minimizer of the @xmath5 minimization @xmath123 is the sparse solution of @xmath124 .
in particular , taking @xmath125 , we obtain @xmath126 , @xmath127 respectively .
it is worth mentioning that @xmath122 is the sharp bound for @xmath4 minimization which has been proved by cai and zhang @xcite . because exact recovery can fail for any @xmath76 $ ]
if the bound of @xmath56 is no less than @xmath128 ( see @xcite ) , @xmath52 is also the sharp bound for @xmath55 minimization . actually , besides @xmath120 , @xmath14 , there are several other @xmath129s satisfying that @xmath92 are integers , for instance @xmath130 , @xmath131 . thus @xmath122 is also a sharp ric bound for such @xmath129s .
\(i ) for any @xmath14 , we can check @xmath132 then from lemma [ lemma4 ] and @xmath133 in theorem [ theorem ] , for @xmath14 and any @xmath76 $ ] , it yields that @xmath134 whose figure ( with @xmath135 ) is plotted as follows .
captypefigure \(ii ) moreover , under some assumptions @xmath136 , since for @xmath137 $ ] @xmath138 and for @xmath139 $ ] @xmath140 then from lemma [ lemma4 ] , we have tab . 2 by calculating limits for cases @xmath141 and @xmath142 of the right - hand side of @xmath143 with @xmath144 .
captypefigure _ tab . 2 : bounds on @xmath145 for any @xmath9 and @xmath12 .
in this section , we state the bound on @xmath15 for any @xmath76 $ ] in the following results : [ theorem1 ] for any @xmath76 $ ] , if _
k < , k2 , + , k3 , holds , then each @xmath3-sparse minimizer of the @xmath5 minimization @xmath113 is the sparse solution of @xmath114 .
particularly , for the case @xmath119 or @xmath92 to be an integer ( such as @xmath120 ) , we have the corollary below by applying lemma [ lemma3 ] . [ prop3 ] for @xmath119 or @xmath88 such that @xmath92 is an integer , if _
k < 1/3 , k2 , + , k3 , hold , then each @xmath3-sparse minimizer of the @xmath5 minimization @xmath113 is the sparse solution of @xmath114 . taking @xmath146 , then @xmath147 from lemma [ lemma3 ] , which produces the bound @xmath18 if @xmath11 is even .
meanwhile @xmath18 for @xmath11 is the sharp bound for @xmath4 minimization that has been gotten by cai and zhang @xcite . from theorem [ theorem1 ] and corollary [ prop3 ] , we list the following table . tab . 3 : upper bounds on @xmath42 for different @xmath90 .
in this paper , we have generalized the upper bounds for rics from @xmath4 minimization to @xmath0 minimization , and established new ric bounds through @xmath5 minimization with @xmath76 $ ] for exact sparse recovery .
an interesting issue which deserves future research would be : how to improve these new bounds for some @xmath76 $ ] when @xmath92 is not an integer .
* proof of lemma [ lemma2 ] * + stimulated by the approach in @xcite , without loss of generality , we only need to prove the case @xmath148 due to the symmetry of components @xmath149 .
clearly , @xmath150 .
notice that if the inequality ( [ l1q ] ) holds for any @xmath151 , then we can immediately generalize the conclusion to all @xmath152 through substituting @xmath153 into ( [ l1q ] ) and eliminating the common factor @xmath154 .
henceforth , it remains to show @xmath155 with @xmath156 , where @xmath79 is a function of @xmath90 specified in ( [ pq ] ) .
first , for any given @xmath76 $ ] define that @xmath157 it is easy to verify that @xmath158 is a convex function on @xmath159 .
since the maximum of a convex function always arrives on the boundary , we have @xmath160\nonumber\end{aligned}\ ] ] letting the distribution of @xmath161 appear for @xmath162 times ( @xmath163 ) in the maximum solution of @xmath164 , we have @xmath165 by the convexity of @xmath166 and @xmath167 , it follows that @xmath168 then it holds that @xmath169 where @xmath79 is defined as ( [ pq ] ) and the last equality holds when @xmath170 $ ] for any @xmath171 $ ] . by computing the first and second order partial derivatives of @xmath79 on @xmath90 , it is easy to verify that @xmath79 is a nonincreasing convex function of @xmath76 $ ] and @xmath172 thus the proof is completed .
* proof of lemma [ lemma3 ] * + if @xmath92 is an integer , then @xmath173 if @xmath92 is not an integer , then @xmath174 due to @xmath175 and @xmath176 , we have @xmath177 now we prove the remaining part @xmath178 .
since @xmath179 and @xmath180 $ ] is a nondecreasing function on @xmath76 $ ] , for any fixed @xmath3 , we can set @xmath181 with sufficient small @xmath182 .
thus @xmath183 it follows readily that @xmath184 the whole proof is finished .
* proof of lemma [ lemma4 ] * + we verify @xmath109 is a nonincreasing function on @xmath185 and a nondecreasing function on @xmath8 . by directly computing the first order partial derivative of @xmath109 on @xmath185 ,
it yields @xmath186 likewise , by computing the first order partial derivative of @xmath109 on @xmath8 , we have @xmath187 then the desired conclusions hold immediately . before proving theorem [ theorem ] , we introduce hereafter several notations . for @xmath188 , we denote hereafter @xmath189 the vector equal to @xmath166 on an index set @xmath190 and zero elsewhere .
especially , we denote @xmath191 as @xmath166 with all but the largest @xmath3 entries in absolute value set to zero , and @xmath192 .
* proof of theorem [ theorem ] * + the approach of this proof is similar as @xcite .
first we consider the case that @xmath193 is an integer . by the null space property @xcite in @xmath5 minimization case , we only need to check for all @xmath194 , @xmath195 suppose on the contrary that there exists @xmath194 , such that @xmath196 .
set @xmath197 and decompose @xmath198 into a sum of vectors @xmath199 , where @xmath200 corresponds to the locations of the @xmath201 largest coefficients of @xmath198 ; @xmath202 to the locations of the @xmath201 largest coefficients of @xmath203 , and so on . that is @xmath204 here
, the sparsity of @xmath205 is at most @xmath201 .
clearly , @xmath206 , which generates @xmath207 . from lemma [ lemma2 ] , for @xmath208 , @xmath209
then we sum @xmath210 for @xmath208 to obtain that @xmath211 we again divide @xmath198 into two parts , @xmath212 , where @xmath213 @xmath214 therefore @xmath215 is @xmath216-sparse as a result of facts that @xmath217 and all non - zero entries of @xmath215 has magnitude larger than @xmath218 .
let @xmath219 , then @xmath220\frac{\alpha}{t-1},\\ \|h^{(2)}\|_\infty&\leq&\frac{\alpha}{t-1}.\end{aligned}\ ] ] applying lemma [ lemma1 ] with @xmath221 , it makes @xmath222 be expressed as a convex combination of sparse vectors : @xmath223 , where @xmath224 is @xmath63-sparse , @xmath225 .
henceforth , @xmath226 for any @xmath227 , denoting @xmath228 , we obtain @xmath229 where @xmath230 are all @xmath231-sparse vectors from the sparsity of @xmath232 , @xmath233 and @xmath234 .
it is easy to check the following identity , @xmath235 since @xmath236 , together with ( [ a ] ) , we have @xmath237 setting @xmath238 , if ( [ dertagk ] ) holds , that is @xmath239 then combining ( [ b ] ) with ( [ bb ] ) , we get @xmath240\nonumber\\ & & -\frac{1-\delta}{4}\sum_{i=1}^{n}\lambda_i(\|h_{max(k)}+ h^{(1)}\|_2 ^ 2+\mu^{2}\|u_i\|_2 ^ 2)\nonumber\\ & = & \sum_{i=1}^{n}\lambda_i[((1+\delta)(\frac{1}{2}-\mu)^2-\frac{1-\delta}{4})\cdot\nonumber\\ & & \|h_{max(k)}+ h^{(1)}\|_2 ^ 2+\frac{1}{2}\delta\mu^{2}\|u_i\|_2 ^ 2]\nonumber\end{aligned}\ ] ] @xmath241\\ & = & \|h_{max(k)}+ h^{(1)}\|_2 ^ 2\cdot\nonumber\\ & & \left[\mu^2-\mu+\delta(\frac{1}{2}-\mu+(1+\frac{g(q , k)}{2k(t-1)})\mu^{2})\right]\nonumber\\ & = & \|h_{max(k)}+ h^{(1)}\|_2 ^ 2(\frac{1}{2}-\mu+(1+\frac{g(q , k)}{2k(t-1)})\mu^{2})\cdot\nonumber\\ & & \left[\delta-\gamma(\mu\left(t,\frac{g(q , k)}{k}),\frac{g(q , k)}{k}\right)\right]\nonumber\\ & < & 0 , \nonumber\end{aligned}\ ] ] where the inequality ( [ c ] ) is derived from the following facts : @xmath242 obviously , this is a contradiction .
when @xmath193 is not an integer , by setting @xmath243 we have @xmath244 and @xmath245 is an integer . utilizing the nondecreasing monotonicity of @xmath109 on @xmath246 for fixed @xmath110 presented in lemma [ lemma4 ] , we can get @xmath247 which can be deduced to the former case .
hence we complete the proof . in order to prove the result theorem [ theorem1 ] , we need another important concept in the rip framework the restricted orthogonal constants ( roc ) proposed in @xcite .
suppose @xmath248 , define the restricted orthogonal constants ( roc ) of order @xmath249 as the smallest non - negative number @xmath250 such that @xmath251 for all @xmath252-sparse vector @xmath253 and @xmath254-sparse vector @xmath255 with disjoint supports .
* proof of theorem [ theorem1 ] * + similar to the proof of theorem [ theorem ] , we only need to check for all @xmath194 , @xmath195 suppose there exists @xmath194 , such that @xmath196 .
set @xmath197 . from the proof of theorem [ theorem
] , we have @xmath256 and @xmath257
. then it follows from lemma 5.1 in @xcite that @xmath258 where the first inequality holds by lemma 5.4 in @xcite and the second inequality by ( [ hold ] ) .
thus from the condition @xmath259 it follows that @xmath260 obviously , this is a contradiction . by lemma 3.1 in @xcite , _ k , k < 2_k , k2 , + _ k k3 .
hence , when @xmath11 is even , it yields that @xmath261 and when @xmath262 is odd , it generates that @xmath263 therefore the theorem is proved .
the work was supported in part by the national basic research program of china ( 2010cb732501 ) , and the national natural science foundation of china ( 11171018 , 71271021 ) . 1 a.m. bruckstein , d.l .
donoho , and a. elad , from sparse solutions of systems of equations to sparse modeling of signals and images .
_ siam rev .
34 - 81 , 2009 .
t. cai , l. wang and g. xu , new bounds for restricted isometry constants , _ ieee trans . inform . theory _ ,
4388 - 4394 , 2010
. t. cai , l. wang and g. xu , shifting inequality and recovery of sparse signals , _ ieee trans .
signal process .
1300 - 1308 , 2010 .
t. cai and a. zhang , sharp rip bound for sparse signal and low - rank matrix recovery , _ appl . and
comput . harmon .
74 - 93 , 2013 .
t. cai and a. zhang , compressed sensing and affine rank minimization under restricted isometry , to appear in _ ieee trans .
signal process .
. t. cai , and a. zhang , sparse representation of a polytope and recovery of sparse signals and low - rank matrices , to appear , 2013 .
cand@xmath264s , j. romberg and t. tao , robust uncertainty principles : exact signal reconstruction from highly incomplete frequency information , _ ieee trans.inf .
489 - 509 , 2006 .
cand@xmath264s and t. tao , decoding by linear programming , _ ieee trans .
inf . theory _ ,
4203 - 4215 , 2005 .
r. chartrand , exact reconstruction of sparse signals via nonconvex minimization , _ ieee signal process .
707 - 710 , 2007 .
x. chen , d. ge , z. wang and y. ye , complexity of unconstrained @xmath265 minimization , to appear in mathematical programming .
x.chen , f. xu and y. ye , lower bound theory of nonzero entries in solutions of @xmath265 minimization , _ siam j. scientific computing _ , vol .
2832 - 2852 , 2010 .
davies and r. gribonval , restricted isometry constants where @xmath266 sparse recovery can fail for @xmath267 , _ ieee trans .
inf . theory _ ,
2203 - 2214 , 2010 .
donoho , compressed sensing , _ ieee trans .
inf . theory _
52 , pp . 1289 - 1306 , 2006 .
eldar and g.kutyniok , compressed sensing : theory and applications , cambridge university press , 2012 .
s. foucart , a note on guaranteed sparse recovery via @xmath5-minimization , _ appl .
_ , vol . 29 , pp .
97 - 103 , 2010 .
s. foucart and m. lai , sparsest solutions of underdetermined linear systems via @xmath5-minimization for @xmath268 , _ appl .
395 - 407 , 2009 .
r. gribonval and m. nielsen , sparse decompositions in unions of bases , _ ieee trans.inf . theory _
3320 - 3325 , 2003 .
lai and j. wang , an unconstrained @xmath5 minimization for sparse solution of under determined linear systems , _
siam j. optimization _ , vol .
82 - 101 , 2011 .
h. rauhut , compressive sensing and structured random matrices , _ radon series comp .
appl . math .
_ , vol . 9 , pp . 1 - 92
y. hsia and r.l .
sheu , on ric bounds of compressed sensing matrices for approximating sparse solutions using @xmath5 quasi norms , to appear , 2012 .
is an associate professor in department of applied mathematics , beijing jiaotong university .
he received his phd degree in operations research from beijing jiaotong university in 2007 . from 2007 to 2009 , he was a post - doctoral fellow of department of combinatorics and optimization , faculty of mathematics , university of waterloo , canada .
his research interests are in sparse optimization , mathematics of operations research .
is a lecturer in the state key laboratory of rail traffic control and safety , beijing jiaotong university .
she received her phd degree in operations research from beijing jiaotong university in 2010 . from 2011 to 2012 , she was a visiting scholar in management science and engineering , school of engineering , stanford university , usa .
her research interests are in sparse optimization , semidefinite programming and interior point methods .
is a professor in department of applied mathematics , beijing jiaotong university .
he received his phd degree in operations research from academy mathematics and system science of the chinese academy of science in 1997 .
he was a research fellow of city university of hong kong from 2000 to 2002 , and he was a visiting scholar of university of waterloo from 2006 to 2007 .
his research interest includes variational analysis , mathematical optimization , mathematics of operations research . | the restricted isometry constants ( rics ) play an important role in exact recovery theory of sparse signals via @xmath0 relaxations in compressed sensing .
recently , cai and zhang @xcite have achieved a sharp bound @xmath1 for @xmath2 to guarantee the exact recovery of @xmath3 sparse signals through the @xmath4 minimization .
this paper aims to establish new rics bounds via @xmath0 relaxation .
based on a key inequality on @xmath5 norm , we show that ( i ) the exact recovery can be succeeded via @xmath6 and @xmath7 minimizations if @xmath1 for any @xmath8 , ( ii)several sufficient conditions can be derived , such as for any @xmath9 , @xmath10 when @xmath11 , for any @xmath12 , @xmath13 when @xmath14 , ( iii ) the bound on @xmath15 is given as well for any @xmath16 , especially for @xmath17 , we obtain @xmath18 when @xmath19 is even or @xmath20 when @xmath21 is odd .
compressed sensing , restricted isometry constant , bound , @xmath5 minimization , exact recovery |
pg 1553 + 113 is a blazar found as part of the palomar - green catalog of uv - excess stellar objects @xcite .
its j2000 coordinates are r.a .
15h55m43.0s , dec .
+ 11d11m24.4s @xcite .
it was classified as a bl lac object due to its featureless optical spectrum @xcite and significant optical variability @xcite .
as occurs in most bl lac objects , the featureless optical spectrum prevents a spectroscopic redshift measurement .
however , several limits have been provided based on indirect measurements ( e.g. * ? ? ?
* ; * ? ? ?
the most recent redshift lower limit was estimated assuming that the host galaxy can be used as a standard candle . for absolute r band magnitudes
mr = -22.5 and mr = -22.9 , @xcite obtains the limits @xmath8 and @xmath9 , respectively .
previously , a more stringent redshift lower limit of @xmath10 was set by @xcite based on the detection of intervening ly@xmath7 absorbers .
this estimation will be used throughout the paper . @xcite also set a redshift upper limit of @xmath11 based on the non - detection of any ly@xmath12 absorbers at @xmath10 .
the vhe @xmath1-ray emission from pg 1553 + 113 was discovered independently and almost simultaneously by h.e.s.s . @xcite and magic @xcite in 2005 . the integral flux recorded by magic at the time of the discovery was @xmath13 above 120gev , and the differential energy spectrum was well described by a power law with a spectral index @xmath14 , compatible with the detection by h.e.s.s .
the source has been monitored with the magic telescopes since 2005 .
the results from the 2005 - 2009 observation campaigns can be found in @xcite .
modest flux variability of a factor of @xmath152.6 on a yearly time - scale has been detected at e@xmath16150 gev , with an integral flux lying in the range from 1.4 to @xmath17 .
the observed energy spectra were well fitted by power laws with photon indices in the range @xmath18 and compatible within uncertainties .
extragalactic vhe @xmath1-rays can be absorbed on the way to the earth via electron - positron pair production when interacting with optical - uv background photons from the extragalactic background light ( ebl , * ? ? ?
* ; * ? ? ?
the ebl is mainly composed of diffuse optical light emitted by stars and partially reprocessed by dust in the ir , redshifted by the expansion of the universe @xcite . the uncertainty on its spectral energy distribution ( sed ) and evolution through the history of the universe still ranges from 20% to 50% at wavelengths 0.4 and 40 microns , respectively .
this uncertainty is mainly due to difficulties in direct measurements . during the past few years
several different approaches have been developed to model the ebl ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and despite the different techniques adopted the resulting ebl models show an overall agreement , differing only marginally .
the @xmath1-ray absorption depends significantly on the energy of the vhe photon , the redshift - dependent sed of the ebl , and the distance to the source . the observed flux ( @xmath19 )
can be expressed as @xmath20 where @xmath21 denotes the intrinsic flux emitted by the source and @xmath22 the ebl optical depth as a function of the energy and redshift .
the ebl imprint on the vhe @xmath1-ray spectrum can be used to set upper limits on the redshift of the source .
this is done by assuming a particular ebl model and a criterion on the intrinsic spectrum , such as a maximum hardness for the reconstructed intrinsic spectrum or the absence of a spectral break with a pile - up at vhe in the reconstructed spectrum .
different authors have used this @xmath1-ray attenuation technique for pg 1553 + 113 , leading to the following limits : @xmath23 @xcite , @xmath24 @xcite , @xmath25 @xcite , @xmath26 @xcite , @xmath27 @xcite , @xmath28 @xcite .
limits on the ebl absorption can be estimated independently from ebl models using the vhe spectrum and the redshift of the source under the assumption that the emission of the source can be properly described by a synchrotron self - compton ( ssc ) model @xcite .
this method has previously been used on pg 1553 + 113 to derive constraints on the @xmath1-ray horizon @xcite .
alternatively , the ebl density relative to that of a given model can be evaluated through the joint analysis of the he or vhe observations of many extragalactic sources , making relatively small assumptions on the shape of the intrinsic spectra @xcite .
pg 1553 + 113 was detected in the high - energy ( he , @xmath29mev ) @xmath1-ray band by the large area telescope ( lat ) on board the _ fermi gamma - ray space telescope _
the energy spectrum for the period 2008 august - 2010 august can be well fitted by a power law with spectral index @xmath30 and f(@xmath31)=@xmath32 and its variability index is 93.5 @xcite .
since the variability index is @xmath16 41.6 the source is variable on a monthly time - scale at @xmath1699% confidence probability .
no flaring activity has been claimed for pg 1553@xmath33113 in the he band to date .
remarkably , a periodic factor - of - two flux enhancement on a monthly time scale over 6 years of fermi operation has been recently found by @xcite .
an extensive multi - wavelength ( mwl ) observation campaign on pg 1553 + 113 was carried out from 2012 february to june , focused on the characterization of its sed as well as the variability of the source emission at different frequencies .
observations from vhe @xmath1-rays to radio were performed : vhe band by magic , he band by _
fermi_-lat , x - rays by _ swift_-xrt , uv - optical observations by _ swift_-uvot , ir by rem and radio by metshovi and ovro . in this paper , the study of the flux variability in the vhe , he and x - ray bands is presented .
the study on the vhe spectrum is focused on the april flare state of the source .
a detailed study on the long - term mwl campaign will be presented in a forthcoming paper .
the paper is organized as follows : section [ data ] describes the mwl data analysis . in section [ results ]
the results are presented .
the light curves from magic , _
fermi_-lat and _ swift_-xrt are shown in section [ lightcurve ] while a detailed analysis on the observed vhe energy spectrum is presented in section [ obs_spec ] .
the intrinsic vhe @xmath1-ray spectrum together with a discussion on ebl imprint and redshift constraints can be found in section [ intrinsic ] .
the sed observed during the flaring state and the theoretical interpretation is described in section [ sed_section ] .
the conclusions can be found in section [ conclusions ] .
the vhe @xmath1-ray observations were performed by the magic telescopes .
the magic system consists of two 17m - diameter imaging atmospheric cherenkov telescopes ( iact ) located on the roque de los muchachos , canary island of la palma ( @xmath34n,@xmath35w ) , at a height of 2200 m above sea level .
the system reaches a sensitivity of @xmath36 of the crab nebula flux for @xmath37gev in 50 h of observations @xcite .
pg 1553 + 113 was observed with the magic telescopes from 2012 february 26 ( mjd 55983 ) to april 26 ( mjd 56043 ) .
the data sample after quality cuts consists of 18.3 h in the zenith angle range @xmath38 to @xmath39 .
the observations were performed in wobble mode @xcite , with the source located 0.4@xmath40 from the centre of the field of view .
the analysis of the data has been performed using the standard magic analysis chain @xcite .
the energy threshold of the analysis is approximately 70 gev .
the source was detected with a high statistical significance ( @xmath41 standard deviations , @xmath42 ) during the time period 2012 february - april .
the emission is compatible with a point - like source at the position of pg 1553 + 113 .
the mean @xmath1-rate during the flare period is @xmath43@xmath1/min for @xmath44 gev .
fermi_-lat is a pair - conversion telescope operating from 20mev to @xmath16 300gev .
further details about the _ fermi_-lat can be found in @xcite .
the lat data reported in this paper were collected from 2012 february 2 ( mjd 55959 ) to june 10 ( mjd 56088 ) . during this period , the _ fermi _ observatory operated almost entirely in survey mode .
the analysis was performed with the ` sciencetools ` software package version v9r32p5 .
the lat data were extracted within a @xmath45 region of interest centred at the location of pg1553@xmath33113 .
only events belonging to the ` source ' class were used .
the time intervals when the rocking angle of the lat was greater than 52@xmath46 were rejected .
in addition , a cut on the zenith angle ( @xmath47 ) was applied to reduce contamination from the earth limb @xmath1-rays , which are produced by cosmic rays interacting with the upper atmosphere .
the spectral analysis was performed with the instrument response functions ` p7rep_source_v15 ` using an unbinned maximum - likelihood method implemented in the science tool ` gtlike ` .
isotropic ( iso_source_v05.txt ) and galactic diffuse emission ( gll_iem_v05_rev1.fit ) components were used to model the background @xcite .
the normalizations of both components in the background model were allowed to vary freely during the spectral fitting .
we evaluated the significance of the @xmath1-ray signal from the source by means of the maximum - likelihood test statistic ts = 2 ( log@xmath48 - log@xmath49 ) , where @xmath50 is the likelihood of the data given the model with ( @xmath48 ) or without ( @xmath49 ) a point source at the position of pg 1553@xmath33113 ( e.g. , * ? ? ?
the model of the region of interest used in ` gtlike ` includes all point sources from the second _ fermi_-lat catalogue ( 2fgl ; * ? ? ?
* ) as well as from a preliminary third _ fermi_-lat catalogue from 4 years of survey observations @xcite that fall within @xmath51 radius around the source .
the spectra of these sources were parametrized by power - law functions , except for 2fglj1504.3.1@xmath331023 , 2fglj1553.5@xmath331255 , and 2fglj1608.5@xmath331029 , for which we used a log - parabola as in the 2fgl catalogue .
a first maximum - likelihood analysis was performed to remove from the model sources having ts @xmath52 10 and/or predicted number of counts based on the fitted model @xmath53 .
a second maximum - likelihood analysis was performed on the updated source model . in the fitting procedure
, the normalization factors and the photon indices of the sources lying within 10@xmath46 of pg1553@xmath33113 were left as free parameters . for the sources located between 10@xmath46 and 15@xmath46 , we kept the normalization and the photon index fixed to the values from the 2fgl catalogue . integrating over the period from 2012 february 2 to june 10 ( mjd 55959 - 56088 ) the fit with a power - law model in the 0.1100 gev energy range results in a ts = 908 , with an integrated average flux of ( 5.7 @xmath54 @xmath55 ) @xmath5610@xmath57
@xmath58 s@xmath59 at the decorrelation energy of 2239 mev and a photon index of @xmath60 = ( 1.59 @xmath54 @xmath61 ) for pg1553@xmath33113 .
using a log - parabola ( lp ) , @xmath62 @xmath63 , the fit yielded for the same period a ts = 910 , with an average flux of ( 4.1 @xmath54 @xmath64 ) @xmath5610@xmath57 ph @xmath58 s@xmath59 , a spectral slope of @xmath7 = 1.49 @xmath54 @xmath65 at the reference energy @xmath66 = 2239 mev , and a curvature parameter @xmath12 = 0.08 @xmath54 @xmath67 .
we used a likelihood ratio test ( lrt ) to check a pl model ( null hypothesis ) against a lp model ( alternative hypothesis ) .
these values may be compared , following @xcite , by defining the curvature test statistic ts@xmath68=(ts@xmath69 - ts@xmath70 ) .
the lrt results in a ts@xmath68 = 2 , corresponding to a @xmath151.4@xmath42 difference .
we , therefore , conclude , that no significant curvature was observed in the lat spectrum of pg 1553@xmath33113 during 2012 february - june due to the poor photon statistics .
similar results were obtained when considering only photons with e@xmath161 gev . above 10 gev
the analysis is strongly affected by the lack of statistics .
no variability during the period was neither found .
however , the fermi - lat spectrum of the source shows curvature when considering a longer integration time interval and we accumulate more photons at the highest fermi - lat energies . in the 3fgl catalog @xcite
, the spectrum of pg1553 + 113 is described by a log - parabola .
the @xmath1-ray light curve using 10-day time bins and a pl model is reported in the middle panel of fig .
[ lc ] . for each time bin , the spectral shape of pg1553@xmath33113 and of all the sources within 10@xmath46 of it
were fixed to the value obtained over the whole period .
the systematic uncertainty in the flux is dominated by the systematic uncertainty in the effective area @xcite . the systematic uncertainty on the effective area amounts to 10% at 100mev , decreasing linearly with the logarithm of energy to 5% between 316mev and 10gev , and increasing linearly with the logarithm of energy up to 10% at 100gev . [ cols="^,^,^,^,^,^,^ " , ] [ ssc_simulation ] .
the gray solid line represents the mean power - law fit of the ten thousand realizations of the toy - mc .
the black circles denote the pg1553 + 113 vhe spectrum ebl - corrected with @xcite model assuming z=0.4 .
the dashed black line represent the best - fitting ssc model from fig .
[ sed].,title="fig : " ]
in this paper we have presented the highest flux state ever detected from the blazar pg 1553 + 113 in vhe @xmath1-rays .
the flare was detected at vhe by the magic telescopes and monitored in he @xmath1-rays by _ fermi_-lat , in x - rays by _ swift_-xrt , in optical - uv by _
swift_-uvot , in infrared by rem , and in radio by metshovi and ovro .
while clear variability has been found in both the vhe and x - ray bands , the he @xmath1-ray flux is compatible with constant emission .
the observed vhe spectrum shows clear curvature and a simple power - law fit is rejected with a confidence level of 4.7@xmath42 .
it is well fitted by a power law with an exponential cut - off or a logparabola .
this curvature most likely originates from ebl absorption if the distance to the source is between the redshift limits measured by @xcite ( @xmath71 ) .
if the real redshift of this source is higher than these limits , the effect would be a hardening of the spectrum or the ( unexpected ) presence of a pile - up in the intrinsic spectrum , which would denote that either the ebl models predict an overestimated ebl level , there is a second emission component at high energies or that more exotic physics needs to be invoked as axion - like particles ( alps ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) or the effect of lorentz invariance violation ( e.g. * ? ? ?
* ) . a redshift upper limit of @xmath72 at 95% c.l . has been derived using the @xmath73 ratio test @xcite .
a quasi - simultaneous sed has been compiled for the flare episode in 2012 april .
it can be well modeled by a one - zone ssc model .
the comparison with previous flux states of the source reveals that the higher frequency part of each sed bump shows higher variability than the lower frequency part .
this fact points to a scenario where the most energetic electrons play a leading role during the flare episodes of the source .
a detailed study of the mwl behaviour and evolution of the sed will be published in a forthcoming paper .
we would like to thank the instituto de astrofsica de canarias for the excellent working conditions at the observatorio del roque de los muchachos in la palma .
the financial support of the german bmbf and mpg , the italian infn and inaf , the swiss national fund snf , the erdf under the spanish mineco , and the japanese jsps and mext is gratefully acknowledged .
this work was also supported by the centro de excelencia severo ochoa sev-2012 - 0234 , cpan csd2007 - 00042 , and multidark csd2009 - 00064 projects of the spanish consolider - ingenio 2010 programme , by grant 268740 of the academy of finland , by the croatian science foundation ( hrzz ) project 09/176 and the university of rijeka project 13.12.1.3.02 , by the dfg collaborative research centers sfb823/c4 and sfb876/c3 , and by the polish mniszw grant 745/n - hess - magic/2010/0 .
part of this work has been possible with the support of the cluster of excellence : `` connecting particles with the cosmos '' , part of the landesexzellenzinitiative hamburg .
the authors thank to s. buson , j. finke , j. perkins and a. neronov for their comments and contributions .
jbg would like to thanks to m. raue for useful discussions .
the _ fermi _ lat collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the lat as well as scientific data analysis .
these include the national aeronautics and space administration and the department of energy in the united states , the commissariat lenergie atomique and the centre national de la recherche scientifique / institut national de physique nuclaire et de physique des particules in france , the agenzia spaziale italiana and the istituto nazionale di fisica nucleare in italy , the ministry of education , culture , sports , science and technology ( mext ) , high energy accelerator research organization ( kek ) and japan aerospace exploration agency ( jaxa ) in japan , and the k. a. wallenberg foundation , the swedish research council and the swedish national space board in sweden .
additional support for science analysis during the operations phase is gratefully acknowledged from the istituto nazionale di astrofisica in italy and the centre national dtudes spatiales in france .
99 abdo a. a. et al . , 2009 , apj , 707 , 1310 abdo a. a. et al .
, 2010a , apj , 708 , 1310 abdo a. a. et al . , 2010b , apjs , 188 , 405 abramowski a. et al . ,
2013a , a&a , 552 , a118 , 14 abramowski a. et al .
, 2013b , a&a , 550 , a4 abramowski a. et al . , 2015 , arxiv:1501.05087 acero f. et al . , 2015 , arxiv:1501.02003 acero f. et al . , 2015 , in prep .
ackermann m. et al . , 2012 , apjs , 203 , 4 ackermann m. et al . , 2012 ,
science , 338 , 1190 aharonian f. et al .
, 2006a , a&a , 448 , l19 aharonian f. et al . , 2006b , nature , 440 , 1018 albert j. et al . , 2007 , apjl , 654 , l119 albert j. et al . , 2008 ,
apj , 674 , 1037 aleksi j. et al .
, 2012a , apj , 748 , 46 aleksi j. et al .
, 2012b , aph , 35 , 435 aleksi j. et al . , 2015 , a&a , 573 , a50 aliu e. et al . , 2015 , apj , 799 , 7 atwood w. b. et al . , 2009 , apj , 697 , 1071 baars j. w. m. et al . , 1977 , a&a , 61 , 99 beasley a. j. et al . , 2002 ,
apjs , 141 , 13 bessell m. s. , castelli f. , plez b. , 1998 , a&a , 333 , 231 biteau j. & williams a. , 2015 , submitted to apj , arxiv:1502.04166 breeveld a. a. et al .
, 2010 , mnras , 406 , 1687 burrows d. n. et al . , 2005 , ssrv , 120,165 cardelli j. a. , clayton g. c. , mathis j. s. , 1989 , apj , 345 , 245 conconi p. et al . , 2004 , in moorwood a. f. m. , masanori i. , eds , proc .
spie , conf .
ser . vol .
5492 , ground - based instrumentation for astronomy .
spie , bellingham , p. 1602
cortina j. et al . , 2012a , astronomer s telegram 3977 cortina j. et al .
, 2012b , astronomer s telegram 4069 covino s. et al .
, 2004 , in moorwood a. f. m. , masanori i. , eds , proc .
spie , conf .
ser . vol .
5492 , ground - based instrumentation for astronomy .
spie , bellingham , p. 1613 danforth c.w .
et al . , 2010 ,
apj , 720 , 976 de angelis a. , roncadelli m. , mansutti o. , 2007 , phys . rev .
d , 76 , 121301 domnguez a. et al .
, 2011 , mnras , 410 , 2556 domnguez a. et al . , 2013 ,
apj , 770 , 77 , 15 pp .
finke j. d. et al . , 2010 , apj , 712 , 1 , 238 - 249 pp .
franceschini a. , rodighiero g. , vaccari m. , 2008 , a&a , 487 , 837 fomin v. p.
et al . , 1994 , apj , 2 , 137 gehrels n. et al . , 2004 ,
apj , 611 , 1005 gilmore r. et al .
, 2012 , mnras , 422 , 3189 - 3207 gould r. j. , schreder g. p. , 1967
, phrv , 155 , 1404 green r.f . , schimdt m. , liebert j. , 1986 , apjs , 61 , 305 hauser m. g. , dwek e. , 2001 , ara&a , 39 , 249 hooper , d. , serpico , p. d. 2007 , phys.rev.lett .
99 , 231102 jacob u. , piran t. , 2008 , phy .
d , 78 , 124010 kalberla p.
m. w. et al . , 2005 , a&a , 440 , 775 knigl , a. , 1981 , apj , 243 , 700 kneiske t. m. , dole h. , 2010 , a&a , 515 , 19 li t .-
, ma y .- q . , 1983 , apj , 272 , 317 liu h. t. , bai j. m. , 2006 , apj , 653 , 1089 lombardi s. et al . , 2011 , in proc .
32nd icrc ( beijing ) ( arxiv:1109.6195 ) mankuzhiyil n. et al . , 2010 , apj 715 l16 maraschi l. , tavecchio f. , 2003 , apj , 593 , 667 massaro e. et al . , 2004 ,
a&a , 413 , 489 mattox j. r. et al . , 1996 ,
apj , 461 , 396 mazin d. , goebel f. , 2007 , apj , 655 , l13 miller h. r. , green r. f. , 1983 , baas , 15 , 957 miller h. r. et al . , 1988 , esa special publication , 281 , 303 moralejo a. et al .
, 2009 , in proc .
31st icrc ( ld ) ( arxiv:0907.0943 ) nolan p. l. et al .
, 2012 , ap.j.supp . , 199 , 31 perlman e. s. et al . , 2005 ,
apj , 625 , 727 prandini e. et al . , 2010 ,
mnras , 405 , l76 prandini e. et al . , 2011 , in proceedings crf2010 ( preprint : arxiv:1101.5005 ) poole t. s. et al . , 2008 ,
mnras , 383 , 627 poutanen j. , zdziarski a. a. , ibragimov a. , 2008 , mnras , 389 , 1427 reimer a. , 2007 , apj , 665 , 1023 richards j. l. et al . , 2011 , apjs , 194 , 29 snchez - conde m. a. et al . , 2009 ,
phy . rev .
d , 79 , i d .
123511 sbarufatti b. , treves a. , falomo r. , 2005 , apj , 635 , 173 sbarufatti b. et al .
, 2006 , aj , 132 , 1 schlafly e. f. , finkbeiner d. p. , 2011
, apj , 737 , 103 schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 scully s. t. , malkan m. a. , stecker f. w. , 2014 , apj , 784 , 138 shaw m. s. et al .
, 2013 , apj , 764 135 sikora m. , moderski r. , madejski g. m. , 2008 , apj , 675 , 71 stecker f. w. , dejager o. c. , salamon m. h. , 1992 , apj , 390 , l49 stecker f. , malkan m. , scully s. , 2006 , apj , 648 , 774 tersranta h. et al .
, 1998 , a&as , 132 , 305 tavecchio f. , mazin d. , 2009 , mnras , 392 , l40 tosti g. et al . , 2004 , in moorwood a. f. m. , masanori i. , eds , proc .
spie conf .
5492 , ground - based instrumentation for astronomy , spie , bellingham , p. 689
tramacere a. et al . , 2007 , a&a , 467 , 501 vaughan s. et al . , 2003 , mnras , 345 , 1271 wilms j. et al . ,
2000 , apj , 542 , 914 zerbi r. m. et al . , 2001 ,
, 322 , 275 @xmath74 ifae , campus uab , e-08193 bellaterra , spain + @xmath75 universit di udine , and infn trieste , i-33100 udine , italy + @xmath76 inaf national institute for astrophysics , i-00136 rome , italy + @xmath77 universit di siena , and infn pisa , i-53100 siena , italy + @xmath78 croatian magic consortium , rudjer boskovic institute , university of rijeka and university of split , hr-10000 zagreb , croatia + @xmath79 max - planck - institut fr physik , d-80805 mnchen , germany + @xmath80 universidad complutense , e-28040 madrid , spain + @xmath81 inst . de astrofsica de canarias , e-38200 la laguna , tenerife , spain + @xmath82 university of d , pl-90236 lodz , poland + @xmath83 deutsches elektronen - synchrotron ( desy ) , d-15738 zeuthen , germany + @xmath84 eth zurich , ch-8093 zurich , switzerland + @xmath85 universitt wrzburg , d-97074 wrzburg , germany + @xmath86 centro de investigaciones energticas , medioambientales y tecnolgicas , e-28040 madrid , spain + @xmath87 institute of space sciences , e-08193 barcelona , spain + @xmath88 universit di padova and infn , i-35131 padova , italy + @xmath89 technische universitt dortmund , d-44221 dortmund , germany + @xmath90 unitat de fsica de les radiacions , departament de fsica , and ceres - ieec , universitat autnoma de barcelona , e-08193 bellaterra , spain + @xmath91 universitat de barcelona , icc , ieec - ub , e-08028 barcelona , spain + @xmath92 japanese magic consortium , division of physics and astronomy , kyoto university , japan + @xmath93 finnish magic consortium , tuorla observatory , university of turku and department of physics , university of oulu , finland + @xmath94 inst . for nucl .
research and nucl .
energy , bg-1784 sofia , bulgaria + @xmath95 universit di pisa , and infn pisa , i-56126 pisa , italy + @xmath96 icrea and institute of space sciences , e-08193 barcelona , spain + @xmath97 universit dellinsubria and infn milano bicocca , como , i-22100 como , italy + @xmath98 now at : nasa goddard space flight center , greenbelt , md 20771 , usa and department of physics and department of astronomy , university of maryland , college park , md 20742 , usa + @xmath99 now at ecole polytechnique fdrale de lausanne ( epfl ) , lausanne , switzerland + @xmath100 now at institut fr astro- und teilchenphysik , leopold - franzens- universitt innsbruck , a-6020 innsbruck , austria + @xmath101 now at finnish centre for astronomy with eso ( finca ) , turku , finland + @xmath102 also at inaf - trieste + @xmath103 also at isdc - science data center for astrophysics , 1290 , versoix ( geneva ) + @xmath104 inaf - ira , i-40129 bologna , italy + @xmath105 aalto university metshovi radio observatory , metshovintie 114 , 02540 , kylml , finland + @xmath106 aalto university department of radio science and engineering , espoo , finland + @xmath107 cahill center for astronomy & astrophysics , caltech , 1200 e. california blvd , pasadena , ca , 91125 , u.s.a .
+ @xmath108 department of physics , purdue university , 525 northwestern ave , west lafayette , in 47907 , usa + [ lastpage ] | pg 1553 + 113 is a very - high - energy ( vhe , @xmath0 ) @xmath1-ray emitter classified as a bl lac object .
its redshift is constrained by intergalactic absorption lines in the range @xmath2 .
the magic telescopes have monitored the source s activity since 2005 . in early 2012 ,
pg 1553 + 113 was found in a high - state , and later , in april of the same year , the source reached its highest vhe flux state detected so far .
simultaneous observations carried out in x - rays during 2012 april show similar flaring behaviour .
in contrast , the @xmath1-ray flux at @xmath3 observed by _
fermi_-lat is compatible with steady emission . in this paper ,
a detailed study of the flaring state is presented .
the vhe spectrum shows clear curvature , being well fitted either by a power law with an exponential cut - off or by a log - parabola .
a simple power - law fit hypothesis for the observed shape of the pg1553 + 113 vhe @xmath1-ray spectrum is rejected with a high significance ( fit probability p=2.6 @xmath4 ) .
the observed curvature is compatible with the extragalactic background light ( ebl ) imprint predicted by current generation ebl models assuming a redshift @xmath5 .
new constraints on the redshift are derived from the vhe spectrum .
these constraints are compatible with previous limits and suggest that the source is most likely located around the optical lower limit , @xmath6 , based on the detection of ly@xmath7 absorption .
finally , we find that the synchrotron self - compton ( ssc ) model gives a satisfactory description of the observed multi - wavelength spectral energy distribution during the flare .
[ firstpage ] gamma rays : observations , blazar , bl lac : agns : individual ( pg 1553 + 113 ) |
mergers of galaxy clusters are the most energetic phenomena in the universe since the big bang . within the context of the hierarchical structure formation scenario ,
cluster mergers are the mechanisms by which galaxy clusters form . in a major merger ,
two subclusters collide at velocities of @xmath4 km s@xmath1 , with a release of gravitational potential energy of @xmath5 ergs .
the gaseous halos of the colliding clusters are subject to hydrodynamical shocks that dissipate a major portion of the released energy by heating the intracluster medium ( icm ) .
cluster merger shocks may also generate turbulence and magnetic fields and may accelerate relativistic particles .
the interaction of cooling cores , which are found in the centers of relatively relaxed galaxy clusters , and merger shocks has been the subject of recent research .
@xcite conducted two - dimensional numerical simulations to study the impact of cluster mergers on the survival of cooling cores for the case of head - on collisions .
they found that for two mass ratios of 16:1 and 4:1 , the crucial parameter that determines the fate of the cooling core of the primary component is the ram pressure experienced by the infalling subcluster .
disruption occurs when a significant mass of the infalling gas reaches the primary s core .
the subcluster gas can heat up the cooling gas and increase the cooling time by a factor of 1040 by processes such as displacement of the cooling core from the potential well , merger shocks and turbulent gas motions .
furthermore , the initial cooling time is important , as the cooling core may reestablish itself , if the final cooling time is less than the hubble time . a time lag of @xmath2 12 gyr was also found between core passage and cooling flow disruption . on the other hand
, @xcite performed numerical simulations to examine the effects of merger shocks on cooling cores .
they concluded that the outer parts of cool cores are being ram pressure stripped , while gas motions , induced inside the core , can result in transport motions in the direction of the incoming shock wave and may eventually lead to the formation of a cold front .
a mechanism to account for cooling flow heating was proposed by @xcite . in their model ,
cluster mergers are thought to produce a significant amount of turbulence in the cluster icm , which in turn generates sound waves that contribute to the heating of the cooling flow and suspend cooling to low temperatures . in this report
, we present results of the x - ray analysis of the cluster abell 2065 .
this is a nearby ( @xmath6 ) richness class 2 cluster @xcite and one of the 10 galaxy clusters that make up the corona borealis supercluster .
two cd galaxies dominate the cluster center in the optical ; their radial velocities differ by @xmath2 600 km s@xmath1 @xcite .
previous x - ray studies with _ asca _ and _ rosat _ led to marginal detections of two surface brightness peaks coincident with the two cd galaxies observed at the cluster center @xcite .
the data indicated that the cluster has undergone a merger .
@xcite suggested that the cooling cores of the two cd galaxies may have survived the merger , and they used this to put constraints on the gravitational potential peaks of the two clusters . the present work is focused on the study of merger signatures that can shed light on the merger dynamics , as well as on the nature of the central cool gas and its association to the two central cds .
the organization of the paper is as follows . in [ sec : data ] we describe the observations and data reduction procedures . in
[ sec : images ] we present x - ray images of the cluster emission and compare them to optical and radio observations .
the x - ray spectral analysis and the derived temperature maps of the cluster are presented in [ sec : spec ] .
the radial x - ray surface brightness profile is presented in
[ sec : sbprof ] and the spectral deprojection is discussed in
[ sec : deproj ] . in
[ sec : disc ] , a possible dynamical interpretation of the merger is proposed .
we summarize our conclusions in [ sec : sum ] . throughout the paper , we use @xmath7 km s@xmath1 mpc@xmath1 , @xmath8 and @xmath9 , which corresponds to a linear scale of 1.336 kpc arcsec@xmath1 at the redshift of the cluster . when errors are reported , they represent 90% confidence intervals .
position angles ( pas ) are measured from north to east .
abell 2065 was observed by _
chandra _ with the acis - i detector on 2002 august 18 for 28 ks ( obs . 1 ) .
the observation was interrupted by a solar flare and resumed on 2002 november 24 for another 22 ks ( obs .
2 ) . the telemetry of the data was performed in very faint mode . due to the considerable pointing and orientation offsets , the initial reduction of the two datasets has been performed independently .
the data have been calibrated using the calibration database , caldb version 3.0 .
the chandra interactive analysis of observations software package , ciao version 3.2 , was used for the reduction and the analysis of the datasets .
the standard filtering has been performed , accepting only events with _ asca _
grades 0,2,3,4 and 6 , and excluding known bad pixels , chips nodes and chip boundaries . in order to reduce the background ,
the extended grades based on the very faint data were used to filter out additional events which were due mainly to particle background .
the data have been corrected for the charge - transfer inefficiency effect ( cti ) .
spatial and temporal corrections for the qe degradation effect were applied automatically by the software .
the acis - s2 chip was used to monitor the background , as it includes less of the cluster emission than the acis - i chips .
the onset of the flare resulted in a high background rate in the august observation ( obs .
time intervals that exhibited background count rate fluctuations greater than 1-@xmath10 from the mean have been excluded , and this resulted in 8 ks being discarded .
even with this clipping of high background data , the mean count rate during the remaining good time is about 30% higher than the expected quiescent value for this period .
however , if the filtering criteria were tightened , nearly all of the time in obs . 1 was rejected . since we are mainly concerned with very bright regions , where very few of the counts are due to background , we retained the resulting 20 ks of obs . 1 , and adjusted the blank sky background files observing length ( exposure keyword in the fits file headers ) for this observation to be consistent with the higher background rate . the background of obs .
2 was much lower .
following the same cleaning procedure as with obs . 1 , less than 2 ks were discarded .
the resultant background rate is @xmath2 1% higher than the quiescent mean for the period of the observation . in order to check the registration of the two observations , point sources
have been detected in each observation using the ciao wavdetectalgorithm .
the positions of the sources were correlated , and the photon positions in the second observation were corrected for the mean offset .
the events lists of the two observations were then merged .
the merged events list was used for all of the analysis , except for spectral fitting .
spectra were extracted separately for the two observations , and were fit by a common model . to check the absolute astrometry of the x - ray data ,
wavdetect was used to detect x - ray point sources in the merged events list .
the x - ray positions of these sources were compared to the positions of optical / ir sources in the 2mass and usno - b1.0 catalogues . a very small shift to the astrometry of the x - ray image was made ( @xmath11 ) in ra and dec , respectively .
the raw x - ray image from the cleaned and merged observations is shown in figure [ fig : raw ] , uncorrected for background or exposure .
the image has been binned so that each pixel occupies 246@xmath12246 on the sky .
this image shows the complex field of view of the combined observations , and the regions where chip gaps complicate the analysis . an adaptively smoothed x - ray image in the 0.310 kev band , corrected for background and exposure variations across the field of view ,
is presented in figure [ fig : smooth ] .
the image was smoothed to a minimum signal - to - noise per smoothing beam of 3-@xmath10 .
overall , the cluster emission is elongated from the northwest to the southeast . at the center , the image shows a compact , bright region . there is a more diffuse tail extending to the north at the center of the image .
although variations in the exposure confuse the region se of the cluster center , there is evidence for a bow - shaped surface brightness discontinuity in this direction .
there is also an extension in the emission further to the se .
the optical dss image of the cluster is presented in figure [ fig : dss]_a _ overlaid with the x - ray contours from the adaptively smoothed x - ray image ( fig .
[ fig : smooth ] ) .
the central cluster region is shown in fig .
[ fig : dss]_b_. the x - ray peak at the cluster center is displaced by 24 from the position of the southern cd galaxy in the cluster . following @xcite , we applied an unsharp masking technique to reveal the structure in our data .
figure [ fig : radio ] shows an image of the central @xmath13 region of abell 2065 produced by this technique .
a 0.310 kev exposure - corrected image was smoothed with two gaussian kernels of 1476 and 1476 ( 3 and 30 pixels on the acis - i detector ) , respectively ; fig .
[ fig : radio ] displays the difference between the two smoothed images .
this particular set of kernels was selected from a range of sets , as they yield the highest degree of structure in the image ; smaller kernels tend to unresolve the largest scale structure , while larger ones tend to wash out the high - resolution features of the image .
accurate positions of the centers of the two cds have been determined from the 2mass catalogue and they are marked with crosses in figure [ fig : radio ] .
the position of the southern cd is close to the brightest region of x - ray emission in the cluster .
the brightest x - rays come from a point just south of the center of the southern cd , and there is an elongated extension of the x - ray emission running about 25 further to the south . there is also a more diffuse tail of emission to the north of the southern cd extending about 45 . the northern cd galaxy is located along the western edge of this tail , but does not correspond to a surface brightness peak or any other interesting feature in the tail . based on much lower resolution _ rosat _ images , @xcite noted that the x - ray emission at the center of abell 2065 was marginally resolved into two peaks ; again , at the lower resolution of the _ rosat _ data , these were consistent with being located at the positions of the two cd galaxies .
@xcite argued that the two peaks were two cool cores associated with separate subclusters with the two cds at their centers , and that these two subclusters had merged .
they suggested that the cool cores were still bound to the two cd galaxies , and drew conclusions about the central concentrations of the dark matter potentials of the two cds . with our higher resolution _
chandra _ data ,
the southern peak is indeed associated with the southern cd , but the northern peak appears to correspond to the more diffuse curved tail .
this is not centered on the northern cd galaxy .
thus , only one of the cds has retained a bound cool core , although the northern cd may not have had a cool core prior to the merger .
the morphology of the northern tail and its apparent connection with the southern cd suggests that the tail may be gas which has been stripped from this cd .
alternatively , it is possible that it was removed from the northern cd , which now retains very little hot gas .
the southern cd galaxy ( 2mass j15222917 + 2742275 ) is also a faint radio source ( 3.09 mjy at 1.4 ghz ) . in figure
[ fig : radio ] , the radio contours from the first survey are superposed on the unsharp - masked x - ray image of the center of the cluster .
the radio source is small and the core is not very well resolved in the first data , although the extent to the south is much larger than the first beam size .
the peak in the radio surface brightness is nearly coincident with the center of the cd galaxy and the peak in the x - ray surface brightness .
the radio source is extended to the se of the center of the cd galaxy .
the radio extension corresponds closely to a similar extension in the x - ray emission .
if both are due to ram pressure as the southern cd moves relative to the surrounding intracluster gas , this would indicate that the galaxy is now moving to the nw .
if this is the case , then the more extended tail to the north is due to gas removed from the northern cd . on the other hand , the radio / x - ray extension to the south
might be due to buoyancy , or to more complex gas motions in the core of the cluster .
we extracted spectra of selected regions in the _ chandra _ image separately for the two datasets .
flux weighted ancillary files and response matrix files were calculated with the ciao tools mkwarf and mkrmf . because in ciao 3.2 the tool mkrmf did not work properly with the latest calibration files ( caldb 3.0 ) , the data were reduced anew using the previous calibration ( caldb 2.8 ) ; the blank sky background files were also processed to match the observations .
the calibration included the temporal and spatial variation of the soft x - ray quantum efficiency degradation due to the build - up of material on the optical blocking filter .
the data were screened using the same good time intervals of low background brightness as in [ sec : data ] .
the spectra for each dataset were independently grouped , so that each bin contained at least 25 counts . in both observations ,
the spectrum at energies below 0.7 kev is affected by sharp effective area and quantum efficiency variations , while at energies above 6.5 kev the background is dominant ; thus we restrict our spectral extraction in this photon energy range 0.76.5 kev .
the spectra from the two datasets were simultaneously fit in this range using the same spectral model .
in most cases , the spectra were fit with the wabs model for the absorption and the apec model @xcite for the thermal emission .
in most fits , the absorbing column density was held fixed to the galactic value of @xmath14 @xmath15 .
the fitting was performed with xspec @xcite , version 11.3 .
first , the spectrum was extracted for the largest elliptical region which roughly followed the shape of the x - ray isophotes and which fell completely within the fields - of - view of both observations ; this was an elliptical region centered on the southern cd galaxy , with semimajor and semiminor axes of 63 and 49 , respectively , aligned with the orientation of the cluster emission ( pa of the major axis at @xmath16 ) .
the background was determined from the blank sky background files discussed in [ sec : data ] .
table [ tab : spec_global ] summarizes the resulting spectral fits . a single temperature model with galactic absorption provided a good fit to the spectrum , with a value of @xmath17 per degree - of - freedom , @xmath18 , very close to unity . the observed spectrum and best - fit model
are shown in figure [ fig : spec_global ] .
we tested the effect of allowing the absorption column density to vary ; this marginally improved the fit , and yielded a column density consistent within the errors with the galactic value ( here and elsewhere , the f - test has been used to compare the fits ) . therefore ,
in the following , we keep the column density fixed at the galactic value .
the best fit model shows a residual structure at @xmath2 2 kev ; it is known that this is generally due to the rapid variation in _
chandra _ s effective area near the ir edge of the telescope s mirror .
we refit the spectrum removing this region ( 1.82.3 kev ) .
this improved the fit , but the model parameters were essentially identical to those including the ir edge region .
in [ sec : spec_tmap ] below , we will see that the cluster contains gas with an extremely wide range of temperatures .
thus , it is somewhat surprising that the global cluster spectrum is well fit by a single temperature model , and so we tried a two temperature model . the fit was improved , but the second thermal component was essentially unconstrained , having a lower limit of about 10 kev .
this indicates that the cluster contains some very hot gas , but the global spectrum is not very useful for characterizing this gas .
previous _ rosat _ and _ asca _ images and spectra suggested that abell 2065 contained a weak cooling flow @xcite .
thus , we also tried a model of a single temperature cluster thermal emission combined with a cooling flow component whose upper temperature was fixed at the cluster ambient temperature and whose lower temperature was 0.08 kev .
the abundance in the cooling flow component was fixed to that of the ambient cluster gas .
xspec s mkcflow , which was used to model the cool core , utilizes the mekal atomic library @xcite to calculate atomic line emission .
in order to use consistent atomic data between the ambient thermal emission and the cool core model components , the mekal model was employed to model the cluster component .
we first tested what effect the use of the mekal model had on the best - fit parameter values as determined with the apec model .
the fit was not significantly improved , and the cluster temperature and abundance were consistent within the errors with the values obtained with apec .
combining the thermal and cooling flow components did not produce a better fit to the spectrum , and the best - fit cluster temperature and abundance values were consistent with those derived from the mekal model alone .
the cooling rate for the model was @xmath2 4@xmath19 m@xmath20 yr@xmath1 , which is in agreement with the measurement of @xcite , but also consistent with no cooling .
the best - fit single temperature model is consistent with results from the _ asca _ spectrum .
@xcite found a best - fit single temperature of @xmath21 kev ; almost exactly the same result was derived by @xcite and @xcite . on the other hand , @xcite derived a temperature of @xmath22 kev from the _ einstein _ mpc detector . both our temperature map ( [ sec : spec_tmap ] , below ) and the previous temperature map from _ asca _ @xcite show that the outer , southern parts of the cluster are very hot .
thus , the higher _ einstein _ mpc average temperature probably results from the very large field of view and a very hard x - ray response of this instrument .
lcccccccc & 852/731=1.166 & @xmath23 & @xmath24 & @xmath25 & & & & + apec & 851/730=1.165 & @xmath26 & @xmath27 & @xmath28 & & & & + apec ir region & 725/661=1.097 & @xmath23 & @xmath29 & @xmath30 & & & & + apec+apec & 817/728=1.122 & @xmath23 & @xmath31 & @xmath32 & & @xmath33 & @xmath34 & + mekal & 839/731=1.148 & @xmath23 & @xmath35 & @xmath36 & & & & + mekal+mkcflow & 839/730=1.149 & @xmath23 & @xmath37 & @xmath36 & & & @xmath38 & @xmath39 since the _ chandra _ x - ray image of abell 2065 shows a wealth of structure and the previous _ asca _ temperature map showed evidence for a shock structure to the south , we have constructed maps of the projected temperature using the interactive spectral interpretation system , isis @xcite .
the temperature maps for two regions centered on the southern cd galaxy have been computed and are presented in figure [ fig : tmaps ] .
[ fig : tmaps]_a _ shows a region that stretches 105 @xmath12 105 across , while the central 41 @xmath12 41 of the cluster is presented in fig .
[ fig : tmaps]_b_. the temperature calculations were performed as follows . at each pixel position ,
the spectrum from each observation was extracted , requiring that at least 800 net counts be included in the energy range of 0.76.5 kev .
the extraction region was a square box centered at the pixel center , whose size varied having a maximum value of 500 acis - i pixels for the larger scale map ( fig .
[ fig : tmaps]_a _ ) and @xmath2 300 acis - i pixels for the cluster center map ( fig . [ fig : tmaps]_b _ ) .
the box size was adaptively determined so that the number of observed counts in the particular region , over the entire energy spectrum , was between 1.5 and 2 times the nominal 800 counts , in order to account for background subtraction as well as for the fact that the fitting occurs over a limited energy range ( 0.76.5 kev ) . for each observation ,
the background was extracted from the blank sky background files , for the same energy band ; for each pixel , the region that was used was the same as the one used with the source spectrum .
the spectra were combined in isis , requiring at least 25 counts per bin , and subsequently fit with a thermal model with galactic absorption .
the abundance was held fixed at the value derived by fitting the global spectrum with a thermal model ( table [ tab : spec_global ] , first row ) . the temperature map of figure [ fig : tmaps]_a _ reveals several spatially separated regions of lower and higher projected temperatures .
the lowest temperatures are found to the nw and in a limited region to the se of the southern cd galaxy .
the entire region nw of the center of the cluster is moderately cool . on the other hand ,
the hottest gas is at radii of @xmath40 3 se of the southern cd .
the typical temperature to the nw of the cluster center is @xmath2 5 kev , while the hotter se region has much higher temperatures of @xmath2 10 kev .
these values are in good agreement with the values @xcite derived from _ asca _ data ( regions # 2 and # 4 , respectively , in their paper ) .
these hot and cool regions seem to be separated by a bow - shaped feature that runs from the ne to the sw of the cd , at @xmath2 3 from it . especially to the se of the cd ,
the temperature gradient is quite steep , increasing by 7 kev over scales smaller than an arcminute in some cases .
the temperature structure of that region and the high temperatures encountered in it , in combination with the bow - shaped feature on the se of the cd suggest the presence of a shock front beyond @xmath2 3 to the se of the southern cd . figure [ fig : tmaps]_c - d _ show , respectively , the negative and positive temperature error at each pixel position , as calculated for 90% error estimates .
the average error in these images is 1.56 kev and 1.70 kev , respectively , for the negative and positive error images .
it is obvious from these images , that the hot region to the ne of the southern cd ( fig .
[ fig : tmaps]_a _ ) is associated with large errors , and thus it could be an artifact of poor statistics . note that @xcite did not find high temperatures in this region ( region # 5 ) with asca . at smaller scales near the cluster center ( fig .
[ fig : tmaps]_b _ ) , the temperature map reveals at least two interesting structures .
the coolest gas in the map seems to be associated with the x - ray extension to the south of the southern cd galaxy , and slightly displaced from the center of this galaxy .
this would be unusual for a cool core in a relaxed system ; generally , they are very accurately coincident with the center of the central cd . as noted previously from fig . [
fig : radio ] , both the thermal x - ray and nonthermal radio plasma near the southern cd appear to be displaced to the south .
the region just north of the southern cd seems to be warmer .
however , the tail further to the north is cooler ( the second coolest region in the cluster ) .
whether the northern tail is actually a distinct gas concentration from the southern extension is not clear from the error images ( fig .
[ fig : tmaps]_c - d _ ) . the warmer material that bridges these two regions has a projected temperature consistent with a monotonic radial increase from the southern extension to the northern tail , but also with a higher temperature than the surrounding regions . in order to address the significance of the apparent temperature increase and to provide a more accurate assessment of the spectral properties of these regions , we have extracted spectra and
fit them with a thermal apec model with galactic absorption .
the geometric regions defined for the extraction purposes were such that they trace these features in the temperature maps ( fig .
[ fig : tmaps ] ) , the raw data image , as well as the unsharp - masked image ( fig . [ fig : radio ] ) .
these regions are displayed in fig .
[ fig : tmaps]_b _ by the solid - line geometrical figures . in detail , the spectrum of the very cool southern extension was extracted from a 12 radius circle centered 101 south of the southern cd center . the spectrum of the warmer region between the two coolest cluster regions ( hereafter , the ` bridge ' ) was extracted from a rectangular region , @xmath2 34@xmath1216 , elongated along the east - west direction and centered 111 to the north of the southern cd .
finally , the spectral extraction for the northern tail was performed from a rectangular region centered 357 to the north of the cd , which was @xmath2 34@xmath1232 in size and run in the east - west direction . since these regions are located at the center of the cluster , their spectra are heavily contaminated by foreground and background x - ray emitting icm along the line - of - sight .
thus , accounting for the background has been accomplished in two different ways : by extracting a local background from the events files themselves and by using the blank sky background files .
the former case has the advantage that foreground / background cluster emission is accounted for in a direct manner , and thus the derived spectral properties should refer to represent the emission of the small cool region near the cluster center .
the dashed - line rectangles in fig .
[ fig : tmaps]_b _ illustrate the regions used for the local background . in detail , the local background for the southern cool core was taken from a rectangular region , @xmath41 , elongated in the east - west direction to avoid emission from the northern structure and centered @xmath2 22 south of the southern cd , excluding , of course , the circular region of the cool core and point sources .
the local background for the bridge was extracted from regions identical to and adjacently positioned to the east and west of the rectangular region used with the spectrum extraction .
for the tail , the background was extracted from 5 rectangular regions , identical and contiguous to the one used for the spectrum extraction ; three of these regions were positioned to the north , and one each to the east and west of the extraction rectangle . on the other hand , when we use the blank sky background spectrum we model the foreground and background emission by including a second thermal component whose temperature and abundance were fixed to the best - fit values from the global cluster x - ray spectrum .
the normalization of this component was allowed to vary .
the best - fit models for the southern cd extension are shown in table [ tab : treg ] .
the two techniques yielded consistent results .
the blank - sky background yielded a better fit , although with higher uncertainty in the best - fit values .
the abundances are essentially unconstrained with both techniques . to test
if the use of the global cluster temperature is a good approximation for the mean icm temperature along the line - of - sight , we fit the spectrum allowing the temperature and abundance of the cluster component to vary .
the fit improved , but the line - of - sight thermal component is now unconstrained , while the uncertainty of the cool core temperature slightly increased .
the abundances are again unconstrained .
consistent results are obtained for the bridge spectrum , as shown in table [ tab : treg ] .
the local background technique yields a better fit and a higher temperature . using the global spectrum temperature to model the cluster emission
did not improve the fit .
this suggests that the global spectrum may not provide an adequate description of cluster emission along this line - of - sight .
thus , we allowed the second thermal component to vary but unfortunately the best - fit values for both thermal components were unconstrained .
this , in combination with the uncertainty in the abundances in both techniques , is a consequence of the low number of counts in this region in both observations .
the two techniques yield consistent results for the northern tail , as shown in table [ tab : treg ] .
the spectrum is adequately fit with either technique , though the blank - sky background technique yielded a better fit .
again , allowing the cluster component to vary slightly improved the fit , but the computed best - fit parameters are more uncertain . as suggested by the temperature map ( fig . [
fig : tmaps ] ) , the northern tail is cooler than the global temperature .
the metallicity is poorly constrained , but it appears to be higher than the cluster average value .
overall , these results suggest that the blank - sky background technique provides slightly better fits , albeit with higher uncertainties . however , the temperature errors that either technique yields are large enough to prevent a definitive assessment of the significance of the temperature increase at the position of the bridge . in
[ sec : unequal ] we argue that the northern tail is a trail of gas that has been stripped from the southern cd galaxy s cooling core . llcccccc south cd extension & local & 63.594/71 = 0.896 & @xmath42 & @xmath43 & & & + south cd extension & blank sky & 58.765/70 = 0.839 & @xmath44 & @xmath45 & & 5.519 & 0.308 + south cd extension & blank sky & 58.281/68 = 0.857 & @xmath46 & @xmath47 & & @xmath48 & @xmath49 + bridge to tail & local & 45.227/59 = 0.767 & @xmath50 & @xmath51 & & & + bridge to tail & blank sky & 53.381/58 = 0.920 & @xmath52 & @xmath53 & & 5.519 &
0.308 + northern tail & local & 96.301/92 = 1.047 & @xmath54 & @xmath55 & & & + northern tail & blank sky & 90.393/91 = 0.993 & @xmath56 & @xmath57 & & 5.519 & 0.308 + northern tail & blank sky & 89.128/89 = 1.001 & @xmath58 & @xmath59 & & @xmath60 & @xmath61
the temperature map ( fig . [ fig : tmaps]_a _ ) shows a complex structure with an inner hot region and an outer very hot region to the se of the southern cd galaxy .
in addition , there are steep surface brightness gradients in the same regions , displayed by the isophotes in fig .
[ fig : dss ] ; these two results suggest the presence of a cold front and a shock front in this region . in order to better understand the gas physics in this cluster region ,
we have extracted the surface brightness profile from a pie circular aperture centered on the southern cd galaxy and lying to the se .
this pie region stretches between 100 and 180 in pa and out to @xmath2 10 from the center of the cd in radius ; a total of 45 annuli were used , their radial extent varying from @xmath2 3 , for the innermost , to @xmath2 30 , for the outermost .
the background was taken from the merged blank sky background file ( [ sec : data ] ) .
the resultant surface brightness distribution is presented in fig .
[ fig : sbprof]_a_. the profile generally shows a rather smooth decline with radius . superposed on that ,
we find two interesting features .
first , there is a sharp decline over the inner part of the profile , terminated by a discontinuity at @xmath2 30from the center .
this discontinuity is associated with the cool southern cd extension .
this indicates that this feature is bounded by a cold front at its southern edge .
additionally , a region of excess x - ray emission , with respect to the outer part of the brightness profile ( see below ) , is found to stretch from @xmath2 30 out to @xmath2 100 .
this is followed by a discontinuity , where the surface brightness drops by a factor of @xmath2 1.7 over a radial extent of 10 , reminiscent of surface brightness jumps observed across shocks , as well as cold fronts .
notice that this discontinuity is not associated with the shock - like features seen in fig .
[ fig : tmaps]_a_. on the contrary , the outer 500 of the profile are consistent with a @xmath62-model surface brightness distribution ( @xmath63 and @xmath64 218 or 292 kpc ) , lacking evident shock signatures .
we have attempted to fit a cold front surface brightness model to the inner part of the profile , but the excess x - ray emission region that extends immediately outwards complicates the analysis .
instead , in an attempt to better study the gas behavior across the entire profile , as well as understand the nature of the features seen in fig .
[ fig : sbprof]_a _ , we have deprojected the surface brightness profile .
the assumptions that underlie this method are those of spherical symmetry with respect to the gravitational potential peak , identified in our case with the southern cd galaxy , and of uniform electron density distribution within each shell .
the electron densities are presented in figure [ fig : sbprof]_b_. the local density distribution clearly demonstrates the two features seen in the projected profile of fig .
[ fig : sbprof]_a_. there is rather rapid central decline until @xmath2 30 , where a discontinuity occurs .
the average density just inside the cold front region is @xmath65 @xmath66 .
then the density undergoes a shallow decline : it drops by a factor of 1.2 over a radial interval of @xmath260 . a dramatic density jump of @xmath22
is observed at @xmath2 100 , followed by a steeper decline .
again , this jump is consistent with being either a shock or a cold front .
the nature of this discontinuity is examined in [ sec : shock ] .
finally , there seems to be another discontinuity at @xmath2 3 from the southern cd , which coincides with the position of the bow - shaped feature of fig .
[ fig : tmaps]_a_. however , this jump also coincides with the region in our merged raw data where the chip gaps of the two observations overlap . the magnitude of the density errors at the position of this jump relative to the errors of adjacent values , in combination with the absence of any strong deviations from the smooth isothermal decline in the surface brightness profile ( fig .
[ fig : sbprof]_a _ ) , indicate that the jump is probably an artifact of low number statistics at the position of the chip gaps .
our interpretation of a shock propagating through the cluster icm has been stimulated by the temperature map structure ( fig .
[ fig : tmaps]_a _ ) .
however , the temperature map does not reflect the local gas temperature ( i.e. the gas at a specific distance from the cluster center ) , rather the average temperature of the x - ray emitting gas along the line - of - sight .
for instance , if a shock is indeed propagating beyond @xmath2 3 se of the southern cd , the projected temperatures of the central cluster region , as shown in the temperature map ( fig .
[ fig : tmaps]_b _ ) , are overestimates of the true local temperatures . in order to disentangle local from projected emission , and gain deeper insight on the gas physics of this region
, we have deprojected the spectra of a set of regions .
specifically , we have defined a set of 8 regions stretching from the center of the optical cd out to a radius of @xmath2 64 and covering the same pa interval as the regions used with the surface brightness deprojection algorithm , namely from 100 to 180 .
these regions have been selected to fall on the acis - i detector in both observations , and in particular to trace features of the raw data image primarily , as well as of the temperature map .
for instance , the innermost region covers most of the area used to examine the spectral properties of the southern cd extension , discussed in [ sec : cool ] , while the third annulus from the center traces the 100brightness discontinuity .
they are displayed in figure [ fig : deprj : reg ] . in these regions ,
the total net counts from both observations range between 1600 and 4200 .
the spectra of these regions were extracted from each observation , as described in
[ sec : spec ] . for each region , the flux weighted sum of the spectra from the two observations for both the source and the background were calculated , and a flux weighted average detector response was produced .
the subsequent fitting was performed in the energy band 0.76.5 kev , grouping bins to contain at least 25 counts , as needed .
the deprojection was carried out with xspec in two different ways ; in both , the shells local emission was modeled with a thermal model with galactic absorption .
first , xspec s mixing model _ projct _ was used .
_ projct _ calculates the projection matrix of each of the overlying shells on the area of each annulus and computes the total projected emission on that annulus .
the fitting occurs simultaneously for all annuli and upon completion the emission integral of the entire spherical shell is returned , along with the best - fitting temperature and abundance values .
fitting all the spectra simultaneously has the effect that the best - fit values of the outer annuli are affected by those of the inner annuli . to reduce the effects of this feature of the algorithm
, we have fit the individual spectrum of the outermost annulus with an apec thermal model with galactic absorption .
then , we used the obtained temperature and abundance values to fix the spectral parameters of the outermost shell during the deprojection and allowing all others to vary we obtained a best - fit @xmath18 value of 0.96 .
the abundance was rather poorly constrained .
an alternate approach was also followed . in this , the spectral fitting occurred successively for the different shells . starting with the outermost spectrum ,
the best - fit values of the spectral parameters were determined .
subsequently , the spectrum of the inner shell was fit with two thermal components , one of which was fixed to the best - fit values of the overlying shell , while the relevant normalization was adjusted to account for projection effects of the outer shell on the inner region and for different exposures between the two regions .
this was done for the spectra of all interior regions , introducing thermal components to account for overlying emission , as needed .
the abundance profile , again , was poorly constrained .
comparing the two methods , it seems that , at least with our data , the latter method yields temperature values with smaller errors .
the profiles presented in the following are based on these results . to check the consistency with the temperature map findings
, we have produced the projected temperature profile for these regions , shown in figure [ fig : tproj ] .
the cool gas at the center of the cluster is followed by constant temperature gas out to a distance of @xmath2 25 from the cd .
the temperature then rises by about 3 kev and maintains a constant value within the errors for @xmath2 2 .
this rise , also seen in the temperature map , has motivated the notion of shock propagation at that region .
subsequently , it drops to values consistent with those derived from _ asca _
data by @xcite .
the temperature map does not show a similar trend at that distance , but this is probably due to poor statistics at the map edge .
although the temperature rise and drop across 4is reminiscent of shock heating , the shock hypothesis is not supported by the detailed x - ray brightness profile ( fig .
[ fig : sbprof ] ) . in the contrary , as was mentioned in
[ sec : sbprof ] , the outer 500 are adequately fit with an isothermal model , and bear no strong evidence for compression . this hot , high entropy gas is then most likely a transient merger - induced effect .
it apparently has a higher pressure than its surroundings and it is expected to dissipate in a few hundred myr ( a few sound crossing times ) . the deprojected temperature profile is presented in figure [ fig : deprj]_a_. the temperature in the central region is quite low , as already indicated in table [ tab : treg ] .
the two high temperature points with very large errors around 3 raise a concern .
if the temperature in these regions has been overestimated , then deprojection might cause the temperatures of inner points to be underestimated ; too much hot projected emission would have been subtracted from these points .
in particular , the apparent temperature drop at about 2 might then be an artifact of the overestimated temperature in the shells just outside . in any case
, the drop at the position of the brightness discontinuity ( at @xmath216 ) is not significant within the errors . given the highly uncertain temperature profile values , we have decided to use the average value of 5.5 kev to describe the gas temperature at the se of the brightness discontinuity .
this value is consistent with temperature of the two outer bins in the profile , and it is also consistent with the the pre - shocked region temperature as estimated from _ asca _
data @xcite . for error propagation purposes we adopt the values obtained by @xcite ( region # 6 of fig
3_c _ , therein ) , that is , @xmath67 kev .
the definition for the thermal model normalization is that : @xmath68 , where z the redshift to the cluster , @xmath69 is the angular diameter distance at that redshift , @xmath70 and @xmath71 the electron and proton number densities , respectively , and the integration is performed over the shell s projected volume . assuming uniform density distribution across each shell , and that @xmath72 , this equation may be inverted to solve for the electron density .
the quality of our data was sufficient to allow calculating accurate errors on the normalization values .
we present the density profile in figure [ fig : deprj]_b_. the profile shows structure similar to the one revealed by the higher resolution density distribution of fig .
[ fig : sbprof]_b_. the density in the central 30 shell is high , and there is an abrupt density drop of about a factor of @xmath2 4 outside of this region . the average density in the central 30 is consistent with the average density derived from the detailed density profile ( fig . [
fig : sbprof]_b _ ) .
another discontinuity is seen at @xmath2 16 , coincident with the position of the brightness discontinuity . a monotonic decline beyond this point
is interrupted by an increase at the outmost shell , where the higher value is an artifact of the assumption inherent to the deprojection algorithm of no external emission . from the temperature and density profiles and assuming the ideal gas equation of state , we have calculated the radial profiles of gas pressure and specific entropy .
they are presented in fig .
[ fig : deprj]_c d _ , respectively .
the entropy was computed using the equation @xmath73 .
note that the quantity plotted in the entropy profile is the residual of the average cluster entropy from the regions entropy , @xmath74 .
the global cluster entropy was calculated from the temperature value of the best - fit model to the global spectrum , and the average cluster density .
note that the cool dense region at the center of the cluster seems to be in pressure equilibrium with its surroundings , while its entropy is the lowest in the cluster .
this suggests that the core is convectively stable . the continuous pressure but
greatly reduced entropy supports the notion that the southern x - ray extension of the southern cd is bounded by a contact discontinuity ( a cold front ) .
the pressure shows a significant drop at @xmath2 16 , which corresponds to the shock feature in fig .
[ fig : sbprof]_a_. there is no entropy decline between the same regions , as would be expected for a shock , although the errors are large .
the outer 2 of the pressure profile are dominated by large errors , but they overall seem to follow a declining trend . similarly , the outer portion of the entropy profile shows rather elevated values , as expected from the projected temperature profile ( [ sec : tproj ] ) .
the detailed electron density profile ( fig . [ fig : sbprof]_b _ ) shows a clear discontinuity at @xmath2 100 ( @xmath2 140 kpc ) to the se of the southern cd .
this discontinuity could be attributed to either a shock front or a cold front at that position .
the critical parameter that would allow one to distinguish between these two possibilities is the value of the temperature contrast across the discontinuity .
the ratio of the temperature for material ahead of the front to the temperature for material behind the front is expected to be greater than unity for a cold front and less than unity for a shock front .
the projected temperature profile bears no evidence for any jump at that location , though the errors are significant . on the other hand , the deprojected temperature profile ( fig . [ fig : deprj]_a _ )
shows a jump consistent with being greater than unity , but , as indicated earlier , it may be an artifact of the deprojection . lacking strong evidence in either direction , we shall examine the implications of both scenarios on the kinematics of the merging system .
first , considering the shock possibility , the density compression across the discontinuity ( @xmath75 ) can be used to constrain the kinematics of the merger . in principle , the mach number @xmath76 of a shock can be determined by employing the rankine - hugoniot shock equations ( @xcite , 85 ) for density or temperature jumps : @xmath77 where the subscript 2 denotes material behind the shock ( the `` post - shock '' region ) , while subscript 1 refers to material ahead of the shock ( the `` pre - shock '' region ) . for intracluster gas , @xmath78 .
using the above compression value with the first part of equation ( [ eqn : shock ] ) , we confirm the presence of a mildly supersonic outflow in this region ( @xmath2 30 - 100 ) with a mach number of @xmath3 at the position of the shock .
the observed temperature jump is much smaller than the theoretical value , @xmath79 , although consistent within the errors .
similarly , the observed and the theoretical values of the pressure ratio are consistent within the errors , although the pressure is not independent of either the density or the temperature .
on the other hand , if the discontinuity is due to a cold front , the velocity can be derived based on the stagnation condition at the cold front @xcite , which gives the ratio of the pressure at the stagnation point at the cold front ( @xmath80 ) to that far upstream ( @xmath81 ) : @xmath82 here , @xmath76 is the mach number of the cold front relative to the upstream gas , and @xmath78 .
we use the density at @xmath2 90 ( fig .
[ fig : sbprof ] ) and the temperature at 12 ( fig .
[ fig : deprj]_a _ ) to approximate the pressure at the stagnation point .
we have chosen to approximate the pressure far upstream with the density value at @xmath2 4 from the cd and the average temperature 5.5 kev ( [ sec : tdeprj ] ) .
then , the pressure contrast is @xmath2 5 , which implies that the cold front is moving through the hotter gas with a mach number of @xmath83 .
this should be viewed as an upper limit to the cold front velocity , as applying equation ( [ eq : pst ] ) for regions closer to the cold front suggests that the motion is still supersonic , but with smaller mach numbers .
it is important to realize that , regardless the ambiguity of the nature of the discontinuity , these two scenarios bear consistent kinematical implications . in the following
, we adopt the mach number derived with the shock condition for the motion of the material at 100 .
the sound velocity for a perfect gas , under adiabatic conditions , is @xmath84 . taking the mean molecular weight of the icm to be @xmath85 , and using the average temperature 5.5 kev ,
the far upstream sound speed is @xmath2 1200 km s@xmath1 .
then , the velocity of the shock or cold front is @xmath86 , or @xmath2 2000 km s@xmath1 , similar to the value derived by @xcite .
the cool core surrounding the southern cd appears to be bounded on its southern edge by a cold front .
an estimate of the merger relative velocity can also be derived based on the stagnation condition at the cold front , equation ( [ eq : pst ] ) . in this case
, @xmath76 is the mach number of the cool core relative to the upstream gas , and @xmath78 . as in the cold front scenario in [ sec : shock ] , the far upstream pressure has been approximated using the density at a distance of @xmath2 4 se of the southern cd and the average temperature of 5.5 kev .
the stagnation point should be in pressure equilibrium with the material inside the cool core ; the density immediately interior to the front ( @xmath87 29 , fig .
[ fig : sbprof]_b _ ) and the deprojected core temperature can be used to approximate the stagnation pressure . for this set of regions ,
the pressure ratio is about 4 , and the mach number is @xmath88 .
this value agrees with the mach number of the discontinuity at 100 . especially in the case
that the latter is indeed a cold front , the discrepancy between the mach number values of the two structures could be understood as being due to acceleration gained by the exterior front , as it propagates down a density gradient .
the cool , low entropy , southern extension has been identified with the cooling flow core of the southern cd . the condition for survival of the cool core is ( roughly ) that the ram pressure force of the incident ambient gas be less than the gravitational restoring force of the potential of the cluster core on the cooling core gas .
assuming that the cooling core was initially in hydrostatic equilibrium , the condition for the survival of the cool core is @xcite : @xmath89 where @xmath90 is the cool core density of the infalling cluster , @xmath91 is the relative velocity , @xmath92 is the pressure at the center of the cool core , and @xmath93 is the ambient pressure of the main cluster around the cool core .
one concern is that this condition assumes that the cooling core gas is bound to the dark matter potential of the main cluster ; in abell 2065 , the center of the cool core is displaced by @xmath2 12 or 16 kpc from the center of the cd .
however , the fact that the cool core remains fairly intact probably implies that this condition is not strongly violated .
the local density at 12 se of the southern cd ( fig . [ fig : sbprof]_b _ ) has been employed for the value of @xmath94 ; the central cooling core pressure @xmath92 was estimated utilizing the deprojected core temperature .
the ambient pressure @xmath95 has been estimated from the density at 4 se of the cd , and the average temperature of 5.5 kev ; it is negligible compared to @xmath92 and is ignored in the following .
the choice for the infalling cluster initial central density is more difficult to make . despite the ambiguity in the nature of the 100 discontinuity and of the high temperature region between 2 and 4
the density immediately outside the 100 front has been taken as a good approximation to @xmath96 .
indeed , even with this uncertainty , equation ( [ eqn : ccsur ] ) yields an upper limit to the relative subcluster velocity of @xmath97 1900 km s@xmath1 , in accordance with the velocity derived from the kinematics of the 100discontinuity .
the results that have been presented above can be compiled into a possible dynamical scenario concerning the history and the stage of the merger . in the following
, we shall assume that the two cds are located at the centers of the two merging subclusters , and that they are physically interacting with each other , ( that is , their close proximity is not a projection effect ) .
we will refer to the cluster centered on the southern cd as the southern cluster , and to the one centered on the northern cd as the northern cluster .
we argue that the southern cluster is more massive than its northern counterpart ; this is corroborated by the evidence presented above and by numerical simulations , as it will be shown below in detail . in this scenario ,
the northern cluster is falling into the more massive southern cluster . with the data
, we can place an upper limit on the density contrast of the two subcluster cores by assuming that the gas ahead of the discontinuity at 100 serves as an undisturbed sample of the initial northern cluster core conditions .
we have considered the density at the position of the x - ray peak ( at @xmath2 2 , fig .
[ fig : sbprof]_b _ ) to be representative of the initial central density of the southern cool core .
then , the original density contrast has an upper limit of 27 .
the absence of gas associated with the northern cd suggests that either the northern cool core has been disrupted during the merger or that the cluster did not possess a cool core originally .
although most subclusters have been observed to possess cool cores , the latter possibility can not be dismissed .
the existence of the cold front at 30 and the discontinuity at 100 to the se of the merger indicate that the southern cd is moving in this direction , having already experienced core crossing .
being the more massive and more dense part of the merger , the southern cluster has swept away most of the northern halo .
the shock that was triggered during the early merger stages has displaced the northern cluster gas halo from the cluster s center of mass and it is now projected to the se of the merger , beyond the discontinuity at 100 . the temperature map structure ( fig . [ fig : tmaps]_a _ ) is similar to the structure revealed in the temperature map of the 1:16 unequal mass merger , with a density contrast of 50 , in the numerical simulations of @xcite , when viewed 250 myr after core crossing .
this indicates that the density contrast may be higher than the upper limit we determined above and that the time since core crossing is probably a few hundred myr .
this also suggests that the cluster has undergone core crossing probably for the first time . in this scenario , the cool core displacement in the direction of motion can be understood in terms of the findings of @xcite
. as the northern cluster was falling into the southern cluster from the se ( before core crossing ) , a shock wave was induced in the southern cluster .
the shock may have reached the dense , cool core of the southern cluster , subjecting the latter to ram pressure stripping , which removed material originally located at the outskirts of the core .
the generated pressure imbalance caused gas motions inside the cool core that resulted in gas transportation against the shock flow towards the contact discontinuity .
finally , in the same scenario , the tail of cool gas that extends from the southern cd to the north can be understood as the material of the core outskirts that was removed by the incoming shock wave .
apparently , this gas did not meet the survival criteria set by equation ( [ eqn : ccsur ] ) , and it has been stripped from the cd . the small temperature difference between the cool core and the cold tail and their similar abundances ( table [ tab : treg ] ) suggest that this may indeed be the case , since the shock could displace as well as heat the gas to the observed values . in [ sec :
images ] the possibility was raised that the southern cd galaxy may be moving to the nw , and that the observed x - ray and radio extensions to the se may be due to ram pressure stripping .
in such a scenario , the two subclusters have not yet achieved closest approach and it is possible that they may be merging for the first time .
however , there are certain issues this possibility fails to address .
the discontinuity to the se of the southern cd galaxy could be a manifestation of a shock wave , generated by the northern cluster as it falls through the icm of the southern cluster .
the induced ram pressure could be responsible for the displacement of the southern cluster cooling flow from the cd .
these considerations in combination with the fact that no prominent shock front appears to the nw in our data , which would be generated by the southern cluster as it free - falls in the potential of its northern counterpart , suggest that the northern cluster must be more massive than the southern .
hence , it is remarkable that the cool core of the northern subcluster does not manifest itself as a region of enhanced x - ray emission in the x - ray brightness map of fig .
[ fig : smooth ] .
additionally , it is much harder to explain the nature of the cool tail that extends to the north of the southern cd , since the numerical experiments of @xcite suggest that a shock front driven into a cooling flow would displace the latter in the incoming shock s direction , but it would not distort its geometry to form an elongated leading structure .
the possibility that this gas may have been expelled from the northern cd is not favored by the temperature consistency between the gas along the tail and the gas of the southern extension .
we have presented the results of a _
chandra _ observation of the merging cluster of galaxies abell 2065 , which include the temperature maps of the central 10 arcmin@xmath98 of the cluster emission and the spectral properties of the cooling flow of the southern cluster as well as a plume of cold gas emerging from the cd to the north .
evidence of shocks appear in the temperature maps to the se of the merger and the deprojected density distribution of that region indicates the presence of a supersonic flow to the se , with @xmath99 1.7
. however , the temperature data are not of sufficient accuracy to allow distinguishing between the shock wave and the cold front interpretation of this discontinuity .
the cooling flow is displaced to the se of the cd . at its southern edge , the cooling core appears to be bounded by a cold front .
putting all these together , we propose that abell 2065 is an unequal mass merger .
the northern cluster seems to have fallen in the more massive southern cluster from the se probably for the first time and having lost its gas content it is now moving to the nw , where it is now seen . we would like to thank the anonymous referee for helpful remarks .
m.c . would like to thank maxim markevitch , adrienne juett and gregory sivakoff for helpful comments and discussions .
support for this work was provided by the national aeronautics and space administration primarily through _ chandra _ award go2 - 3159x , but also through go2 - 3160x , go3 - 4160x , go4 - 5133x , go4 - 5137x , and go5 - 6126x , issued by the chandra x - ray observatory , which is operated by the smithsonian astrophysical observatory for and on behalf of nasa under contract nas8 - 39073 .
houck , j. c. , denicola , l. a. 2000 , in asp conference proceedings , astronomical data analysis software and systems ix , ed .
nadine manset , christian veillet , & dennis crabtree , ( san francisco : asp ) , 591 | we present an analysis of a 41 ks _ chandra _ observation of the merging cluster abell 2065 with the acis - i detector .
previous observations with _
rosat _ and _ asca _ provided evidence for an ongoing merger , but also suggested that there were two surviving cooling cores , which were associated with the two cd galaxies in the center of the cluster .
the _
chandra _
observation reveals only one x - ray surface brightness peak , which is associated with the more luminous , southern cd galaxy . the gas related with that peak is cool and displaced slightly from the position of the cd .
the data suggest that this cool material has formed a cold front . on the other hand , in the higher spatial resolution _
chandra _
image , the second feature to the north is not associated with the northern cd ; rather , it appears to be a trail of gas behind the main cd .
we argue that only one of the two cooling cores has survived the merger , although it is possible that the northern cd may not have possessed a cool core prior to the merger .
we use the cool core survival to constrain the kinematics of the merger and we find an upper limit of @xmath0 1900 km s@xmath1 for the merger relative velocity .
a surface brightness discontinuity is found at @xmath2 140 kpc from the southern cd ; the mach number for this feature is @xmath3 , although its nature ( shock or cold front ) is not clear from the data .
we argue that abell 2065 is an example of an unequal mass merger .
the more massive southern cluster has driven a shock into the icm of the infalling northern cluster , which has disrupted the cool core of the latter , if one existed originally .
we estimate that core crossing occurred a few hundred myr ago , probably for the first time . |
soon after the discovery of superconductivity in lafeaso@xmath5f@xmath6 , the superconductivity was also detected in the binary compound fese@xmath5 @xcite with transition temperature @xmath7k .
this compound possesses the simplest crystal structure among the new families of fe - based superconductors , and consists of a stack of fe square planar layers , which are tetrahedrally coordinated by se atoms . also the large pressure effect on transition temperature
was later observed @xcite with @xmath8k at pressures @xmath9gpa , indicating that fese@xmath5 is actually a high temperature superconductor .
therefore , the superconducting fese@xmath5 compound has attracted considerable attention and is a subject of intensive studies for the last years @xcite .
the structural simplicity of fese favors experimental and theoretical studies of chemical substitution and high pressure effects , which are aimed at promoting a better understanding of a mechanism of the superconductivity . upon cooling below room temperature ,
the tetragonal @xmath10 phase of fese@xmath5 undergoes a subtle distortion to the lower symmetry orthorhombic @xmath11 phase @xcite .
this transition occurs within a broad temperature range , about 70@xmath12100 k , depending on the stoichiometry of the samples .
it was also found that the tetragonal phase undergoes structural transitions under high pressures ( @xmath13 gpa ) to the hexagonal non - superconducting @xmath14 nias - type phase @xcite , and then to its orthorhombic modification ( @xmath15 , mnp - type ) @xcite .
the recent theoretical examination of stability regions of the high pressure phases of fese@xmath5 @xcite indicated a possibility of metallization and superconductivity in the orthorhombic @xmath15 phase under high pressure .
though a substantial increase of @xmath16 was clearly observed under pressure @xcite , these studies did not detect any magnetic ordering .
however , recent nmr studies provided some indication of an incipient magnetic phase transition under pressure @xcite .
recently , a static magnetic ordering has been detected above @xmath17gpa by means of zero - field muon spin rotation ( zf @xmath18sr ) @xcite and neutron diffraction @xcite .
these studies have revealed that as soon as magnetic ordering emerges , the magnetic and superconducting states apparently coexist , and both the magnetic ordering temperature @xmath19 and @xmath16 increase simultaneously with increasing pressure . also , it was recently found , that upon applying pressure the increase of @xmath16 in fese@xmath5 appeared to be nonmonotonic and exhibits a local maximum at @xmath20 0.8 gpa , which is followed by a local minimum at @xmath21 1.2 gpa @xcite .
therefore , there is still a considerable controversy regarding an interplay between electronic structure , magnetism and superconductivity in the fese@xmath5 compound , especially under pressure .
the experimental data on magnetic susceptibility of fese@xmath5 system in the normal state are incomplete and contradicting @xcite .
also , these data are mostly obtained on polycrystalline samples and often distorted by the presence of secondary magnetic phases of iron . in order to shed more light on the relation between magnetic and superconducting properties ,
it is very important to elucidate the intrinsic susceptibility of fese@xmath5 superconductors and investigate its evolution with doping , temperature , and pressure .
here we report on results of the experimental studies of magnetic properties for single - crystalline and polycrystalline fese samples of high quality in the normal state .
these studies include measurements of the temperature dependence of magnetic susceptibility and its anisotropy as well as the hydrostatic pressure effects .
the experimental investigations are supplemented by _ ab initio _ calculations of the electronic structure and magnetic susceptibility of fese .
the calculations are based on the local - spin - density approximation ( lsda ) of the density - functional theory ( dft ) .
the results of experiments and calculations are used to analyze the nature of magnetism in fese and the basic mechanisms of its strong pressure dependence .
the plate - like single crystals of fese@xmath5 superconductor were grown in evacuated quartz ampoules using the kcl / alcl@xmath22 flux technique with a constant temperature gradient of 5@xmath23 c per cm along the ampoule length ( temperature of the hot end was kept at 427@xmath23 c , temperature of the cold end was about 380@xmath23 c ) .
typical dimensions of the produced single crystalline samples are ( 2@xmath123)@xmath24(2@xmath123)@xmath24([email protected] ) mm@xmath25 .
the tetragonal @xmath10 structure was demonstrated at room temperature by x - ray diffraction technique . the energy dispersive x - ray spectroscopy , performed on a cameca sx100 ( 15 kev ) analytical scanning electron microscope , revealed fe : se=1:[email protected] composition denoted in the following as fese for simplicity .
the study of magnetic properties of fese samples at ambient pressure was carried out at @xmath27 k by using a squid magnetometer .
the superconducting transition was detected within 6@xmath128 k. the magnetization dependencies @xmath28 in magnetic fields up to 5 t appeared to be close to linear , indicating that the concentrations of ferromagnetic impurities are negligibly small .
a typical temperature dependence of the magnetic susceptibility for single crystalline fese samples is shown in fig . [ x(t ) ] .
as is seen , a substantial growth of susceptibility with temperature was revealed in the normal state , as well as a large magnetic anisotropy .
temperature dependence of the magnetic susceptibility in the normal state for the single - crystalline fese measured in the field @xmath29 mt .
data corresponding to the magnetic field directions @xmath30 axis and @xmath31 are denoted by @xmath32 and @xmath33 symbols , respectively . ]
pressure dependencies of the magnetic susceptibility , normalized to its value at @xmath34 , for the polycrystalline fese compound at temperatures 78 and 300 k. the solid lines are guides for the eye . ]
the study of magnetic susceptibility of fese under helium gas pressure @xmath35 up to 0.2 gpa was performed at fixed temperatures 78 and 300 k by using a pendulum - type magnetometer placed directly in the nonmagnetic pressure cell @xcite . in order to measure the pressure effect with a reasonable accuracy
, a sufficiently large mass of the sample is required .
we have used the fese sample , further called as `` polycrystalline '' fese , which was prepared by compacting a number of about 50 of small arbitrarily oriented single crystals inside of an aluminum foil cylinder .
the total mass of the sample was about 200 mg .
the measurements were carried out in the field @xmath36 t and their relative errors did not exceed 0.5% .
the experimental pressure dependencies @xmath37 at different temperatures are shown in fig . [ x(p ) ] , which demonstrates their linear character .
the negative value of the pressure effect , dln@xmath0/d@xmath38 gpa@xmath39 , was observed at temperature 300 k , whereas at @xmath40 k the effect appeared to be positive , dln@xmath0/d@xmath41 gpa@xmath39 .
the available experimental and theoretical results on dln@xmath0/d@xmath35 for fese are compiled in table [ dlnx / dp ] together with corresponding data for the relative fete compound for comparison . .[dlnx
/ dp ] pressure derivatives of the magnetic susceptibility dln@xmath0/d@xmath35 ( in units 10@xmath42 gpa@xmath39 ) for fese and fete compounds at different temperatures . [ cols="^,^,^,^",options="header " , ] @xmath43 from nmr knight
shift data of ref .
+ @xmath44 results for fete are taken from ref .
in order to analyze the magnetic properties of fese compound in the normal state , the _ ab initio _ calculations of the electronic structure and paramagnetic susceptibility were carried out . at ambient conditions fese compound possesses the tetragonal pbo - type crystal structure ( space group @xmath10 ) , which is composed by alternating triple - layer slabs .
each iron layer is sandwiched between two nearest - neighbor layers of se , which form edge - shared tetrahedra around the iron sites .
the positions of selenium layers are fixed by the structural parameter @xmath4 , which represents the relative height of se atoms above the iron plane .
the structural parameters of fese were determined by means of x - ray and neutron diffraction @xcite .
the main purpose of the present _ ab initio _ calculations was to evaluate the paramagnetic response in an external magnetic field and to elucidate the nature and features of magnetism in the normal state of fese compound . in the context of this task ,
the dependencies of the magnetic susceptibility on volume , lattice parameters and temperature were addressed . _
ab initio _ calculations of the electronic structure of fese were performed by employing a full - potential all - electron relativistic linear muffin - tin orbital method ( fp - lmto , code rspt @xcite ) .
the exchange - correlation potential was treated within the local spin density approximation @xcite of the dft .
the calculated basic features of electronic structure of fese appeared to be in a qualitative agreement with results of previous dft calculations @xcite .
band structure of fese around the fermi level ( at 0 ev ) marked by a horizontal line . ] as is seen in figs .
[ fese_bs ] and [ fese_dos ] , the calculated band structure and density of electronic states ( dos ) of fese indicate the presence of the hybridized predominantly @xmath45-like fe electronic states close to the fermi level @xmath46 .
the chalcogen @xmath47-states are situated well below @xmath46 and slightly hybridized with the @xmath45-states of iron .
also , one can see a van hove singularity in fig .
[ fese_dos ] at @xmath48 mev below @xmath46 .
it should be noted , that a proximity of the van hove singularity to the fermi level is considered as the key ingredient for superconductivity in iron - based superconductors @xcite .
total density of electronic states of fese ( solid line ) and the partial contribution of the iron @xmath45-states ( dashed line ) .
the fermi level position at 0 ev is marked by a vertical line . ]
calculated paramagnetic susceptibility of fese as a function of the unit cell volume ( in @xmath25 ) .
@xmath4 is taken to be 0.26 .
the arrows indicate the theoretical ( 1 ) and experimental ( 2 ) equilibrium volume values . ] calculated paramagnetic susceptibility of fese as a function of internal parameter @xmath4 for the experimental unit cell volume .
the arrow indicates the experimental value of @xmath4 . ] to evaluate the paramagnetic susceptibility of fese , the fp - lmto - lsda calculations of field - induced spin and orbital ( van vleck ) magnetic moments were carried out within the approach described in ref . .
the relativistic effects , including spin - orbit coupling , were incorporated , and the effect of an external magnetic field @xmath49 was taken into account self - consistently by means of the zeeman term : @xmath50 here @xmath51 is the bohr magneton , @xmath52 and @xmath53 are the spin and orbital angular momentum operators , respectively . the field induced spin and orbital magnetic moments provide estimations of the related contributions to the magnetic susceptibility , @xmath54 and @xmath55 .
for the tetragonal crystal structure the components of these contributions , @xmath56 and @xmath57 , are derived from the magnetic moments calculated in the external field of 10 t , which was applied parallel and perpendicular to the @xmath58 axis , respectively .
it is found that magnetic response to the external field is very sensitive to the unit cell volume , as well as to the structural parameter @xmath4 , which represents the relative height of chalcogen species from the fe plane .
the calculated dependencies of susceptibility of fese as functions of the volume and parameter @xmath4 are given in figs .
[ x(v ) ] and [ x(z ) ] , respectively . calculated temperature dependence of the paramagnetic susceptibility of fese .
@xmath4 is taken to be 0.26 , the unit cell volume is fixed between the theoretical and experimental values .
the dashed line is a guide for the eye . ] also , the thermal effects are taken into account in order to calculate the temperature dependence of paramagnetic susceptibility for fese compound . in this case
the field induced spin and orbital magnetic moments were evaluated by corresponding integration with the energy derivative of the fermi - dirac distribution function @xmath59 , and the temperature dependence of @xmath0 was actually determined by taking into account the finite width of @xmath60 ( see ref . for details ) .
it should be noted that the energy derivative of the fermi - dirac distribution corresponds to a dirac delta function at low temperatures , having a sharp positive peak at the fermi energy @xmath46 .
this steep behavior of @xmath60 resulted in some instability in the numerical calculations of @xmath0 , which are seen in fig .
the experimental superconducting transition temperatures 6@xmath128 k , obtained for the studied fese@xmath5 samples , agree with those reported in literature @xcite . above @xmath16 , a substantial growth of susceptibility with temperature
is revealed in the normal state of fese up to 300 k ( fig .
[ x(t ) ] ) , what indicates the itinerant nature of electronic states of fe at the fermi energy .
the total susceptibility in the absence of spontaneous magnetic ordering can be represented as the sum : @xmath61 where these terms correspond to the pauli spin susceptibility , a generalization of the van vleck orbital paramagnetism , the langevin diamagnetism of closed ion shells , and a generalization of landau conduction electrons diamagnetism , respectively .
obviously , the @xmath62 term does not provide any anisotropy , and for fese the langevin diamagnetism of closed ion shells can be estimated according to ref . as @xmath63 emu / mol . in order to analyse the experimental data on @xmath64 for fese we used the calculated paramagnetic contributions to susceptibility , @xmath54 and @xmath55 .
it has been shown @xcite that for paramagnetic metallic systems the stoner approach underestimates the spin susceptibility , whereas the lsda field - induced calculations take into account non - uniform induced magnetization density in the unit cell and provide more adequate description .
the first - principles calculations of the paramagnetic susceptibility for fese revealed that this system is in close proximity to the magnetic critical point .
this can be seen from a steep rise of @xmath65 and @xmath66 in figs . [ x(v ) ] and [ x(z ) ] , respectively , above the corresponding experimental values of @xmath67 and @xmath4 .
in fact , the calculated stoner enhancement @xmath68 clearly indicates that fese is close to the ferromagnetic stoner instability . for the unit cell volume chosen between the theoretical and experimental values , the dominant spin contribution to magnetic susceptibility of fese
is estimated to be @xmath69 emu / mol .
the averaged orbital @xmath55 term amounts to @xmath70 emu / mol ) , being of about 15% of the total paramagnetism . from comparison of the calculated paramagnetic susceptibility for the ground state in fig .
[ xtt ] , @xmath71 emu / mol , with the experimental value @xmath72 at @xmath73 k in fig .
[ x(t ) ] , @xmath74 emu / mol , it is clear that calculated value @xmath75 has to be substantially compensated by a diamagnetic contribution in order to conform with the experimental data .
one can estimate the expected diamagnetic contribution to magnetic susceptibility of fese to be about @xmath76 emu / mol .
this diamagnetism is comparable in absolute value with the paramagnetic contribution , being much larger than the estimated above langevin diamagnetism @xmath62 .
apparently , it can be ascribed to the @xmath77 term in eq .
( [ chitot ] ) . according to the experimental data in fig .
[ x(t ) ] , the observed anisotropy of susceptibility @xmath3 is large in fese , and even comparable with the averaged susceptibility itself at low temperatures .
it appears to be much larger than the calculated anisotropy of orbital contribution to paramagnetic susceptibility of fese , @xmath78 emu / mol .
therefore , in order to explain the experimental @xmath3 one can assume the presence of a substantial and presumably anisotropic diamagnetic contribution from conduction electrons . to calculate the landau diamagnetic contribution
@xmath77 is a rather difficult problem @xcite .
the free - electron landau approximation , which is often used for estimations , provides @xmath79 that equals @xmath80 of the pauli spin susceptibility .
however , for many metallic systems the diamagnetism of conduction electrons @xmath77 can be many times larger than the free - electron landau estimate @xmath81 , and such anomalous and anisotropic diamagnetism is often determined by the presence of quasi - degenerated states with small effective masses at @xmath46 ( see ref . and references therein ) .
as can be seen in fig .
[ fese_bs ] , in fese the quasi - degenerate states with small effective masses exist at @xmath46 around the symmetry points @xmath82 and z , where the band degeneracies are lifted by the spin - orbital coupling .
such band structure features are of particular importance in connection with a manifestation of the anomalously large and anisotropic @xmath77 , which was found to originate from the similar degeneracy points @xcite .
it should be emphasized that rigorous theoretical analysis of @xmath77 is a rather cumbersome procedure , which goes beyond the aims of the present work . at this stage , we have identified appropriate electronic states near @xmath46 as possible sources of the large and anisotropic conduction electrons diamagnetism in fese . the theoretically evaluated temperature dependence of the paramagnetic susceptibility @xmath83 in fig
. [ xtt ] , which takes into account the finite width of the energy derivative of the fermi - dirac distribution function , actually provides only a slight increase in @xmath0 with temperature .
thus , the observed substantial growth of @xmath84 is rather puzzling at present .
it is presumably related to a fine structure of dos at @xmath46 , but one should expect that fese system is driven far from the ground state at room temperatures . at this stage we should admit that increase of the unit cell volume for the tetragonal @xmath10 phase ( @xmath85 with temperature rising up to 300 k , see ref . ) can provide about 10% growth of paramagnetic susceptibility , according to the @xmath65 dependence in fig .
[ x(v ) ] .
however , this volume expansion does not explain the experimental @xmath84 in fig .
[ x(t ) ] .
also , a change of the structural parameter @xmath4 with temperature can be substantial and of importance due to the strong @xmath66 dependence in fig .
[ x(z ) ] , but to date the influence of temperature on @xmath4 was not studied in a systematic way .
therefore , noticeable temperature effects on the lattice parameters , chalcogen atom position @xmath4 , and electronic structure itself should be taken into account in a rigorous quantitative analysis of @xmath84 in fese .
the measured pressure effects on magnetic susceptibility of fese are intriguing and require a detailed examination .
firstly , as can be seen in table [ dlnx / dp ] , there is a striking sign difference for the pressure effects on @xmath0 at low and room temperatures . also , the absolute value of this effect is substantially larger than that observed in strongly enhanced itinerant paramagnets @xcite , and appeared to be comparable with such pressure effect on @xmath0 reported for the related fete compound @xcite ( see table [ dlnx / dp ] ) .
temperature dependencies of the nmr knight shift @xmath86 in fese measured at ambient pressure ( @xmath32 ) and at @xmath87 gpa ( dashed line ) .
the data are taken from ref . .
the inset shows dependence of @xmath86 on the averaged magnetic susceptibility for fese , @xmath88 , from fig . [ x(t ) ] . ]
it should be noted that the present experimental data on @xmath89 for fese are in reasonable agreement with the results of ref . on temperature and pressure dependencies of the nmr knight
shift @xmath86 of fese in the normal state . as can be seen in fig .
[ nmr ] , the temperature dependence @xmath90 at ambient pressure reflects the corresponding dependence of the magnetic susceptibility in fig . [ x(t ) ] .
assuming the latter to be governed by the spin susceptibility @xmath91 , the only temperature dependent contribution in @xmath90 can be determined as @xmath92 with @xmath93 % ( emu / mol)@xmath39 , resulted from the slope of @xmath94 linear dependence in inset of fig .
in addition , this contribution is expected to be also responsible for the pressure effect on @xmath86 . using a rough approximation @xmath95
we obtain an estimate of dln@xmath0/d@xmath96 dln@xmath97/d@xmath35 , which is listed in table [ dlnx / dp ] .
pressure dependence of the internal chalcogen structural parameter @xmath4 for fese .
the solid line indicates the results of calculations from ref . .
experimental data on parameter @xmath4 in fese for the tetragonal phase at @xmath98 k ( @xmath33 , ref . ) , @xmath99 k ( @xmath100 , ref . ) , @xmath101 k ( @xmath102 , ref . ) , and for the orthorhombic phase at @xmath103 k ( @xmath32 , ref . ) and @xmath104 k ( @xmath105 , ref . ) . in inset : data of ref . on the expanded scale . ] in order to clarify the behavior of @xmath106 in fese , we carried out the _ ab initio _ calculations of paramagnetic susceptibility as a function of pressure .
these calculations are based upon the pressure dependent structural parameters , which have been calculated and listed in ref . .
as is seen in fig .
[ z(p ) ] , the calculated behavior of @xmath107 follows the available experimental data @xcite . in the course of corresponding calculations of @xmath106 for fese we evaluated the pressure derivative dln@xmath0/d@xmath108 gpa@xmath39 in the range 0@xmath121 gpa .
the evaluated derivative appeared to be in a qualitative agreement with the experimental low temperature data ( see table [ dlnx / dp ] ) .
as part of these electronic structure calculations for fese , we also obtained a pressure dependence of the density of states at the fermi level , which is presented in fig .
[ dos(p ) ] .
in addition to the calculated structural parameters from ref . , we also employed the small upward shift @xmath109 to start the optimized @xmath107 dependence from the experimental value @xmath110 ( see @xmath107 behavior in fig .
[ z(p ) ] ) .
the corresponding two sets of calculations demonstrate in fig .
[ dos(p ) ] a tolerable variation of @xmath111 behaviors in fese under pressure , depending on a small adjustment of @xmath112 between the theoretical and experimental values . with the aim to elucidate the main mechanism of the experimentally observed strong increase of the magnetic susceptibility of fese under pressure at low temperatures
, we have also analyzed the pressure effect in terms of the corresponding changes of the volume and @xmath4 parameters by using the relation : @xmath113 the required values of the partial volume and @xmath4 derivatives of @xmath0 can be estimated from the results of _ ab initio _ calculations presented in figs .
[ x(v ) ] and [ x(z ) ] , and were found to be @xmath114 and @xmath115 for the values of @xmath67 and @xmath4 close to the experimental data .
the optimized value dln@xmath67/d@xmath116 gpa@xmath39 is taken for the compressibility of fese @xcite .
this calculated compressibility agrees closely with the experimental values @xmath117 gpa@xmath39 @xcite .
also , the optimized value d@xmath4/d@xmath118 gpa@xmath39 was adopted for evaluation of eq .
( [ dx / dv ] ) .
as is seen in fig .
[ z(p ) ] , this value of d@xmath4/d@xmath35 is in agreement with the experimental data at low pressures .
as far as all parameters entering eq .
( [ dx / dv ] ) are estimated , the first term in ( [ dx / dv ] ) results in a large negative value @xmath119 gpa@xmath39 , whereas the second term appears to be large and positive : @xmath120 gpa@xmath39 . the both terms in eq .
( [ dx / dv ] ) taken together yield the estimate dln@xmath0/d@xmath121
gpa@xmath39 for fese , which is consistent with the low temperature experimental data .
this estimate is also close to the _ ab initio _ calculated pressure derivative based on theoretical lattice parameters from ref .
( dln@xmath0/d@xmath108 gpa@xmath39 , see table [ dlnx / dp ] ) .
actually , the difference in these evaluated values of dln@xmath0/d@xmath35 , depending on whether experimental or theoretical lattice parameters are employed , covers a reasonable range for the expected pressure effect on @xmath0 .
based on the results of calculations , presented in figs .
[ x(v ) ] and [ x(z ) ] , the observed hydrostatic pressure effect on @xmath0 in fese at low temperatures can be represented as a sum of two large in size and competing contributions , related to the pressure dependence of the structural parameters @xmath67 and @xmath4 . as a result
, the experimental positive value of the pressure effect , dln@xmath0/d@xmath41 gpa@xmath39 , is determined by a dominating contribution from the change of @xmath4 under pressure . calculated pressure dependencies of the density of states at the fermi level for fese.the pressure dependent structural parameters , including the lattice constants and chalcogen atom position @xmath4 , were taken from the calculations of ref .
( @xmath32 ) . for another set of @xmath111 calculations ( @xmath100 ) the small upward shift @xmath109
was employed to start from the experimental value @xmath110 ( see @xmath107 behavior in fig .
[ z(p ) ] ) .
the solid lines are guides for the eye . ]
actually , the nature of the large positive pressure effect on @xmath0 in fese is similar to that reported for fete compound @xcite .
however , in the case of fete such effect is twice more pronounced , and also takes place at room temperatures , whereas for fese dln@xmath0/d@xmath35 is found to be negative at 300 k ( see table [ dlnx / dp ] ) .
the reason of this difference is unclear and have to be elucidated . at the present stage one can presume , that the negative sign of dln@xmath0/d@xmath35 derivative is probably related to a nature of the observed anomalous growth of @xmath84 up to room temperatures ( fig .
[ x(t ) ] ) , which is not the case for fete @xcite .
it appears that at higher temperatures this anomalous growth of @xmath84 in fese is apparently reduced by applied pressure . basically ,
the observed positive pressure effect on @xmath0 in fese at low temperatures correlates with the calculated increase of the density of states at the fermi level @xmath111 at low pressures ( see fig .
[ dos(p ) ] ) . at higher pressures one can see nonmonotonic variation of @xmath111 in fig .
[ dos(p ) ] which clearly exhibits consecutive maximum and minimum .
it was recently shown @xcite , that superconducting transition temperatures @xmath16 of a number of iron - based superconductors correlate with the corresponding values of the density of states at the fermi level , thus supporting the bcs - like pairing mechanism in these systems . remarkably , that the presently calculated behavior of @xmath111 under pressure ( the upper curve in fig .
[ dos(p ) ] , with maximum at 0.5 gpa and minimum at 2.2 gpa ) is qualitatively consistent with the reported experimental dependencies of @xmath122 in fese ( corresponding maximum and minimum of @xmath122 were observed at @xmath123 gpa and @xmath124 gpa , respectively @xcite ) .
the calculated pressure dependence of dos at the fermi level for fese with the structural parameters taken from ref .
( the lower curve in fig . [ dos(p ) ] ) also contains consecutive maximum and minimum of @xmath111 , which are substantially shifted to higher pressures .
the magnetic susceptibility of fese compound is found to rise substantially with temperature , which apparently points to the itinerant nature of the electronic states of fe .
the calculated paramagnetic susceptibility @xmath83 describes qualitatively the experimental dependence @xmath125 , however the origin of the observed about twofold increase of @xmath0 up to 300 k is puzzling . from comparison of the experimental values of susceptibility and its anisotropy
with that calculated for paramagnetic contributions to @xmath0 in fese , the additional anisotropic diamagnetism is expected to be of the order of @xmath126 emu / mol , which can relate to the diamagnetism of conduction electrons and presumably has its origin in the quasi - degenerate electronic states close to @xmath46 . the measurements of magnetic susceptibility under hydrostatic pressure revealed a strong positive effect at low temperatures .
this effect appeared to be comparable with that reported for the related fete compound @xcite , whereas at room temperature the pressure effect for fese is found to be also strong , but _
negative_. our calculations indicate that paramagnetic susceptibility of the fese compound is substantially dependent on the unit cell volume @xmath67 and the relative height @xmath4 of se species above the fe plane .
it is shown that the observed at low temperatures large positive pressure effect on @xmath0 is related to the strong sensitivity of the paramagnetic susceptibility to the parameter @xmath4 , which determines the dominant positive contribution .
the grounds of the negative sign of dln@xmath0/d@xmath35 derivative in fese at 300 k are unclear and probably linked to a nature of the observed anomalous growth of @xmath84 . at present
one can state that at higher temperatures this anomalous growth of @xmath84 in fese is apparently reduced by applied pressure .
the present _ ab initio _ calculations have demonstrated that for fese compound the nonmonotonic behavior of superconducting transition temperature with pressure qualitatively correlates with the density of electronic states at the fermi level .
this indicates a possible realization of the bcs - like pairing mechanism in this system .
this work was supported by the russian - ukrainian rfbr - nasu project 01 - 02 - 12 and 12 - 02 - 90405 , by russian ministry of science and education through state contracts 11.519.11.6012 and 14.740.11.1365 , by nasu young scientists grant 03 - 2012 , and by a grant of the president of the russian federation for state support of young russian scientists ( mk-1557.2011.5 ) .
99 f. c. hsu , j. y. luo , k. w. yeh , t. k. chen , t. w. huang , p. m. wu , y. c. lee , y. l. huang , y. y. chu , d. c. yan , and m. k. wu , proc .
38 * , 14262 ( 2008 ) .
s. medvedev , t. m. mcqueen , i. a. troyan , t. palasyuk , m. i. eremets , r. j. cava , s. naghavi , f. casper , v. ksenofontov , g. wortmann , and c. felser , nature materials * 8 * , 630 ( 2009 ) .
s. margadonna , y. takabayashi , y. ohishi , y. mizuguchi , y. takano , t. kagayama , t. nakagawa , m. takata , and k. prassides , phys .
b * 80 * , 064506 ( 2009 ) .
d. braithwaite , b. salce , g. lapertot , f. bourdarot , c. marin , d. aoki , and m. hanfland , j. phys .
matter * 21 * , 232202 ( 2009 ) .
t. m. mcqueen , q. huang , v. ksenofontov , c. felser , q. xu , h. zandbergen , y. s. hor , j. allred , a. j. williams , d. qu , j. checkelsky , n. p. ong , and r. j. cava , phys .
b * 79 * , 014522 ( 2009 ) .
e. pomjakushina , k. conder , v. pomjakushin , m. bendele , and r. khasanov , phys .
b * 80 * , 024517 ( 2009 ) . j. n. millican , d. phelan , e. l. thomas , j. b. leao , and e. carpenter , solid state commun . * 149 * , 707 ( 2009 ) . y. mizuguchi and y. takano , j. phys
. soc . jpn . *
79 * , 102001 ( 2010 ) .
r. hu , h. lei , m. abeykoon , e. s. bozin , s. j. l. billinge , j. b. warren , t. siegrist , and c. petrovic , phys .
b * 83 * , 224502 ( 2011 ) . j .- y .
lin , y. s. hsieh , d. a. chareev , a. n. vasiliev , y. parsons , and h. d. yang , phys .
b * 84 * , 220507(r ) ( 2011 ) .
a. v. fedorchenko , g. e. grechnev , v. a. desnenko , a. s. panfilov , s. l. gnatchenko , v. v. tsurkan , j. deisenhofer , h .- a .
krug von nidda , a. loidl , d. a. chareev , o. s. volkova , and a. n. vasiliev , fiz .
nizk . temp . *
37 * , 100 ( 2011 ) [ low temp . phys . *
37 * , 83 ( 2011 ) ] . y. mizuguchi and y. takano , z. kristallogr . *
226 * , 417 ( 2011 ) .
g. garbarino , a. sow , p. lejay , a. sulpice , p. toulemonde , m. mezouar , and m. nunez - regueiro , europhys .
* 86 * , 27001 ( 2009 ) .
r. s. kumar , y. zhang , s. sinogeikin , y. xiao , s. kumar , p. chow , a. l. cornelius , and c. chen , j. phys .
b * 114 * , 12597 ( 2010 ) .
s. s. naghavi , s. chadov , and c. felser , j. phys .
matter * 23 * , 205601 ( 2011 ) .
a. v. fedorchenko , g. e. grechnev , v. a. desnenko , a. s. panfilov , s. l. gnatchenko , v. tsurkan , j. deisenhofer , a. loidl , o. s. volkova , and a. n. vasiliev , j. phys .
matter * 23 * , 325701 ( 2011 ) .
j. m. wills , m. alouani , p. andersson , a. delin , o. eriksson , and a. grechnev , _ full - potential electronic structure method : energy and force calculations with density functional and dynamical mean field theory _ ( springer , berlin , 2010 ) . | a detailed magnetization study for the novel fese superconductor is carried out to investigate the behavior of the intrinsic magnetic susceptibility @xmath0 in the normal state with temperature and under hydrostatic pressure .
the temperature dependencies of @xmath0 and its anisotropy @xmath1 are measured for fese single crystals in the temperature range @xmath2 k , and a substantial growth of susceptibility with temperature is revealed .
the observed anisotropy @xmath3 is very large and comparable with the averaged susceptibility at low temperatures . for a polycrystalline sample of fese , a significant pressure effect on @xmath0
is determined to be essentially dependent on temperature . _
ab initio _ calculations of the pressure dependent electronic structure and magnetic susceptibility indicate that fese is close to magnetic instability with dominating enhanced spin paramagnetism .
the calculated paramagnetic susceptibility exhibits a strong dependence on the unit cell volume and especially on the height @xmath4 of chalcogen species from the fe plane .
the change of @xmath4 under pressure determines a large positive pressure effect on @xmath0 which is observed at low temperatures .
it is shown that the literature experimental data on the strong and nonmonotonic pressure dependence of the superconducting transition temperature in fese correlate qualitatively with calculated behavior of the density of electronic states at the fermi level . |
we report on a comparative study of the valence band electronic structure of nio as bulk material and of nio as impurity in mgo . from the impurity
we have been able to determine reliably the parameters which describe the local correlations , thereby establishing the compensated - spin character of the first ionization state or the state created by hole doping .
using bulk - sensitive x - ray photoemission we identify pronounced satellite features in the valence band of bulk nio which can not be explained by single - site many body approaches nor by mean field calculations .
we infer the presence of screening processes involving local quasi - core states in the valence band and non - local coherent many body states .
these processes are strong and the propagation of an extra hole in the valence band of nio will therefore be accompanied by a range of high energy excitations .
this in turn will make the observation of the dispersion relations in the ni @xmath0 bands difficult , also because the effective band width is no more than 0.25 ev as estimated from multi - site calculations .
nio is a benchmark system in solid state physics .
it crystallizes in the nacl structure , has a partially filled @xmath0 shell ( ni@xmath1 @xmath2 ) , and is an antiferromagnetic insulator with a nel temperature of 523 k @xcite .
it was pointed out early on by de boer and verwey @xcite that many of the properties of the @xmath0 transition metal compounds do not agree with the predictions of band theory , e.g. , standard band theory predicts nio to be metallic .
a qualitative explanation was proposed in terms of the mott - hubbard model @xcite in which the on - site ni @xmath0-@xmath0 coulomb interaction plays a decisive role .
an early _ ab initio _ attempt to fix the shortcoming of band theory was to treat nio as a slater insulator in which the doubling of the unit cell allows for the existence of a gap @xcite .
however , the calculated gap of about 0.2 ev @xcite turned out to be much too small : a combined photoemission ( pes ) and bremsstrahlung - isochromat ( bis ) spectroscopy study showed that the band gap is 4.3 ev @xcite and established thereby the correlated nature of nio .
the inclusion of a self - interaction - correction ( sic ) or hubbard @xmath3 term to the density - functional formalism may provide a justification for the magnitude of the experimental band gap @xcite . yet , one of the most direct methods to critically test the accuracy of the different approaches , is to determine the excitation spectrum associated with the introduction of an extra particle into the system @xcite .
curve ( a ) in fig .
1 displays the valence band x - ray photoemission spectrum ( xps , @xmath4 = 1486.6 ev ) of an _ in situ _ cleaved nio single crystal .
this spectrum represents essentially the ni @xmath0 spectral weight since the photoionization cross section of the o @xmath5 is relatively small @xcite .
one can clearly observe from curve ( e ) in fig . 1 that the ni @xmath0 density of states calculated by band theory ( in the local density approximation , lda ) does not match at all
: it has a fermi cut - off and the line shape is completely different .
the inclusion of the hubbard @xmath3 in the calculations ( lda+u ) does not solve the line shape problem , see curve ( d ) .
all this demonstrates the shortcomings of mean field theories to describe spectra associated with the fundamental one - particle green s function of the system @xcite .
a completely different approach is to give up the translational symmetry of the system in order to focus on the local correlations and , especially , the dynamics of the propagation of the injected particle . curve ( c ) of fig .
1 shows the ni @xmath0 spectral weight from an early cluster configuration - interaction calculation by fujimori and minami @xcite , which also includes the full atomic multiplet theory .
the agreement with the experimental spectrum is extremely good
. nevertheless , a later cluster calculation by van elp _ et al . _
@xcite arrived at a less satisfactory result : peak b has almost disappeared in the calculation , see curve ( b ) . the prime motivation to use a different set of model parameters here is to infer that the first ionization state is low spin ( @xmath6 ) @xcite rather than the hund s rule high spin ( @xmath7 ) , analogous to the case of zhang - rice singlets in the cuprates @xcite .
recent developments combining lda with dynamical mean field @xcite or gw approaches @xcite yield ni @xmath0 spectral weights which deviate in important details from the experimental spectrum .
these discrepancies between the experiment and the later theoretical simulations @xcite do not provide confidence that one has made progress in understanding the nature of the first ionization state . the issues that we need to address now are threefold .
first of all we have to establish whether the xps valence band spectrum in fig .
1 is truly representative for bulk nio .
there are reports in the literature claiming that certain satellite peaks in the ni @xmath5 spectrum are due to surface effects @xcite .
second , we have to determine to what extent a single - site many body approach can be utilized to describe the electronic structure of nio for which band formation is also essential .
third , we need to identify the nature of the first ionization state in the framework of a local _ ansatz_. to this end we measured the valence band of nio utilizing the more bulk - sensitive hard x - ray photoelectron spectroscopy ( haxpes ) and we investigated experimentally the electronic structure of nio impurities in mgo . the xps data ( @xmath8 ev ) on _ in situ _ cleaved nio single crystals were recorded using a vacuum generators twin crystal monochromator al-@xmath9 source and an scienta ses-100 analyzer , with an overall energy resolution set to 0.35 ev .
the haxpes data ( @xmath10 ev ) were taken at the taiwan beamline bl12xu of spring-8 in hyogo , japan using an mbs a-1he analyzer .
the overall energy resolution was set to 0.35 ev .
the nio impurity in mgo system was prepared _ in situ _ as 10 - 20 nm thin films on polycrystalline ag by means of molecular beam epitaxy .
the measurements were performed at the 11a1 dragon beamline of the nsrrc in hsinchu , taiwan .
the photoemission spectra were recorded at the cooper minimum of ag @xmath11 ( @xmath12 ev ) @xcite using a scienta ses-100 analyzer with an overall energy resolution set at about 0.15 ev . in fig .
[ fig2 ] we show the valence band photoemission spectra of a freshly cleaved nio bulk crystal , taken with a photon energy of 1486.6 ev ( xps ) and 6500 ev ( haxpes ) . by increasing the photon energy
we increase also the kinetic energy of the outgoing photoelectron and , thus , also the inelastic mean free path .
one can estimate that the probing depth is then enhanced from about 15 to roughly 80 @xcite .
we observe that the spectra are very similar .
we , thus , conclude that the xps data as displayed in figs . 1 and 2 is representative for the nio bulk material and that the contribution of surface effects @xcite can be safely neglected . to be specific : peak b is intrinsic for bulk nio .
we would like to note that increasing the photon energy from 1486.6 ev to 6500 ev does not alter much the ni @xmath0 character of the spectrum .
the o @xmath5 photoionization cross section relative to that of the ni @xmath0 remains very small , it changes from 1/13 to only 1/10 @xcite , meaning that peak b truly belongs to the ni @xmath0 spectral weight and not to the o @xmath5 @xcite .
mg@xmath13o and an mgo reference , together with the resulting difference spectrum . also included is the result of a single - site nio@xmath14 configuration - interaction cluster calculation .
the inset shows the ni @xmath15 x - ray absorption spectrum of the ni@xmath16mg@xmath13o together with that of bulk nio . ]
the valence band spectrum of the ni@xmath16mg@xmath13o impurity system is shown in fig .
[ fig3 ] together with the spectrum of an mgo reference thin film grown simultaneously under identical oxygen and substrate conditions .
the ni@xmath16mg@xmath13o film ( and also the mgo film ) was capped by 2 monolayers of mgo in order to prevent the surface termination to have an effect on the local electronic structure of the ni impurity .
the inset in the figure displays the ni @xmath15 x - ray absorption spectra of the ni@xmath16mg@xmath13o and the nio bulk .
the spectra are essentially identical , verifying that the ni in the ni@xmath16mg@xmath13o has very similar local surrounding ( nio@xmath14 octahedra ) as in the bulk .
the valence band spectra of the ni@xmath16mg@xmath13o and mgo systems are normalized to their o @xmath17 core level intensities .
both are dominated by the o @xmath5 valence band , yet , there are clear differences between them due to the presence or absence of the 5% nio impurity . the difference spectrum multiplied by a factor of 6
is given by the red curve in fig .
[ fig3 ] .
the line shape remains the same for films with lower ni concentrations , as is the case for that of the ni @xmath5 @xcite .
this curve represents essentially the ni @xmath0 spectral weight of the nio impurity since the photoionization cross section of the ni @xmath0 is an order of magnitude larger than that of the o @xmath5 at the photon energy used @xcite .
remarkable is that it is different from the spectrum of bulk nio as shown in figs . 1 and 2 .
the impurity spectrum lacks specifically peak b which is prominently present in the bulk spectrum . to interpret and understand the impurity spectrum ,
we have performed simulations using the well - proven configuration - interaction cluster model which includes the full atomic multiplet theory @xcite .
the simulations have been carried out for a nio@xmath14 cluster using the program xtls 8.3 @xcite .
the bottom curve in fig .
[ fig3 ] shows the ni @xmath0 one - electron removal spectrum from the cluster calculation .
the agreement with the experiment is very satisfactory . in order to achieve this ,
we have started the calculations by using parameter values which were suggested from earlier studies on nio @xcite .
we then fine - tune the parameters describing the octahedral crystal and ligand fields , and also the difference between the hubbard @xmath3 and the o @xmath5-ni @xmath0 charge transfer energy @xcite .
the crucial issue here is to obtain a main line ( peak a ) without having another feature appearing at about 2 ev higher energies ( peak b ) as was the case in the simulations by fujimori and minami @xcite and by van elp _ et al .
_ @xcite .
this has implications for the energetics of the states making up the valence band as we explain in the following . a detailed look at the cluster calculations displayed in fig .
1 shows that peak a is given by the @xmath18 final state of the ni 3@xmath19 multiplet structure while peak b is due to the @xmath20 . avoiding the appearance of peak b means that the energy splitting between these two states must be made smaller , e.g. , 1 ev or less .
this is what we have done in our simulation in fig .
3 , using different but equally reasonable parameter values @xcite .
the consequences for the physics are , yet , quite far reaching .
given the fact that various x - ray absorption studies find an effective octahedral crystal and ligand field splitting of about 1.65 ev @xcite , i.e. , the splitting between the isospin @xmath20 and @xmath6 states , we arrive at the conclusion that the @xmath6 must be lower in energy than the @xmath18 by 0.65 ev or more .
this is what we read from our results in fig .
3 . in other words ,
our impurity study provides the spectroscopic evidence that the first ionization state has a compensated - spin character rather than the hund s rule high - spin .
this in turn justifies that the ground state of a hole doped nio system may indeed be low - spin in nature @xcite . and o @xmath5 spectral weights of the nio valence band from a lda+dmft calculation ( reproduced from ref .
@xcite ) and the result of a three - site ni@xmath21o@xmath22 configuration - interaction cluster calculation . ]
we now return to the problem of the bulk nio valence band spectrum . fig .
[ fig4 ] shows the ni @xmath0 spectral weight taken with xps and compares it with the spectra of the nio impurity and of the single - site lda+dmft calculation @xcite .
one can clearly observe that peak b is absent in the impurity as well as in the ni @xmath0 spectral weight of the single - site calculation .
in fact , one could infer that the calculation reproduces quite well the impurity spectrum , with perhaps some discrepancies due to the incomplete implementation of the multiplet structure of the on - site coulomb interactions .
yet , the discrepancy with the bulk spectrum strongly suggests that the origin of peak b must be sought in non - local correlations , i.e. , effects which can not be included in a single - site approach .
our suggestion is that peak b is due to non - local screening processes involving the formation of low - energetic coherent many body states on neighboring nio clusters , which are of the @xmath6 type as we have shown above .
the mechanism is analogous as proposed earlier for the ni @xmath5 core level spectrum of bulk nio @xcite , but the application of it for the valence band is only valid for local states which are relatively stable against band formation .
this may not be applicable for the @xmath6 state , which is a state in which a hole is injected in an @xmath23 orbital starting from the @xmath24 @xmath25 ground state @xcite .
this hole can be expected to readily propagate in the lattice since the hopping between the ni @xmath0(@xmath23 ) and o @xmath5(@xmath26 ) orbitals are rather large @xcite , yet , it may leave behind an energetically costly wake of wrong spins in the antiferromagnetic lattice . in any case
, it would not be meaningful to describe its band formation as a low - energy screening process involving neighboring @xmath6 states @xcite .
hole state by the photoemission process .
( b ) screening by a next nearest neighbor nio@xmath14 cluster producing a coherent low - energetic @xmath6 hole state there .
] however , for the main peak of the bulk nio spectrum , i.e. , the @xmath18 state , we infer that we can make a meaningful approximation by using the coherent @xmath6 screening model .
the @xmath18 consists of a hole injected into the @xmath27 orbital , and its ability to move is rather limited since the overlap between the ni @xmath0(@xmath27 ) and o @xmath5(@xmath28 ) is small .
one could consider the @xmath18 as a localized quasi - core state .
we then can invoke the non - local screening process as follows : after the creation of the @xmath18 state , an @xmath23 electron from a neighboring nio cluster hops onto the ni site , leaving behind a coherent @xmath6 hole state on that neighbor . a sketch for this process
is given in fig .
[ fig5 ] .
these two states are energetically almost degenerate @xcite , and the ni @xmath0(@xmath23 ) and o @xmath5(@xmath26 ) hybridization between them is then strong enough to produce two peaks : not only the main peak a but also the satellite peak b. to confirm our assignments , we have performed a ni@xmath21o@xmath22 cluster calculation consisting of three edge - shared nio@xmath14 octahedra . while all the o 2@xmath29 and ni
3@xmath30 orbitals are included for the nio@xmath14 octahedron in the center where the photo - excitation takes place , those on the other parts of the cluster are replaced by a reduced basis set using the method in ref .
@xcite .
the results are displayed in fig .
[ fig4 ] and demonstrate the presence of both peaks a and b. note that we have used the same parameters as for the single - site calculation which produces only peak a , see fig .
[ fig3 ] .
we also note that the energy difference between peaks a and b is somewhat smaller and the intensity of peak b is slightly larger than those of the experiments .
this can be explained by the fact that the number of neighboring ni sites is only two in the ni@xmath21o@xmath22 cluster : the energy difference will increase and the intensity of peak b will decrease for a larger number of neighboring sites @xcite . to summarize : we have succeeded to determine reliably the ni @xmath0 valence band spectra representative for bulk nio as well as for nio as an impurity system . from the impurity data we are able to extract the local electronic structure and the correlations herein , thereby establishing firmly the compensated - spin character of the first ionization state . by comparing the bulk with the impurity system
, we were able to identify features in the bulk nio spectrum which are caused by screening processes involving local quasi - core valence band states and non - local low - energetic many body states .
these processes are strong and cause considerable redistribution of the spectral weight when an extra hole is injected into nio .
this explains why fine features related to the dispersion of the ni @xmath0 band is difficult to detect @xcite , also since the effective band width is only of order 0.25 ev as estimated from our multi - site cluster calculations .
we gratefully acknowledge the nsrrc and the spring-8 staff for providing us with beamtime .
we would like to thank l. hamdan for her skillful technical and organizational assistance .
the research in germany is partially supported by the deutsche forschungsgemeinschaft through sfb 608 and for 1346 . |
|
@xcite and @xcite have proposed an empirical star formation law between the surface densities of the star formation rate ( @xmath7 ) and the total gas ( @xmath8 ) , widely known as the kennicutt - schmidt ( k s ) law .
s law is written in the form of @xmath9 as stars are formed from molecular gas instead of atomic gas , @xmath8 in the k
s law is often replaced with @xmath5 . with @xmath0co observations and the galactic standard co - to - h@xmath1 conversion factor ( @xmath4 @xmath10 2.3 @xmath11 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 ; @xcite ) ,
numerous works show that @xmath6 is in the range of 1.0 2.0 .
the slope of the k s law may provide us with clues that will help to explain the mechanism of star formation . for example , if star formation only depends on the mass of giant molecular clouds ( gmcs ) , a constant sfr per unit mass of molecular gas is expected , leading to a linear relation between @xmath15 and @xmath5 and , namely , @xmath16 @xcite .
if molecular gas is converted to stars with a timescale of free - fall time , and the scale height of gas is approximately constant everywhere in a galaxy , a slope of 1.5 is expected @xcite .
if cloud - cloud collisions trigger star formation , the star formation timescale is set by the collision rate of the virially bound objects , resulting in a slope of @xmath172.0 @xcite .
there are , however , some drawbacks of the current method of deriving the k s law . at first , @xmath4 is not a constant but a function of metallicity and co intensity @xcite .
observations have also shown that @xmath4 increases with decreasing oxygen abundance @xcite and co intensity @xcite .
because oxygen abundance appears to be a function of the galactic radius in spiral galaxies @xcite , @xmath4 may vary within a single galaxy as well .
for example , @xmath4 is 0.75 @xmath11 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 within the inner 1@xmath182 of m51 .
it is approximately three times lower than the galactic standard @xmath4 , while its value rises to ( 1 4 ) @xmath11 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 at @xmath19 2@xmath180 @xcite .
for this reason , applying the galactic standard @xmath4 to external galaxies is not ideal .
secondly , the nature of @xmath0co also limits its ability to determine @xmath5 .
@xmath0co is optically thick , so it traces the emissions from the envelope of the gmcs , rather than the dense clumps / cores in the gmcs , where stars are formed . likewise , because of its high opacity , the intensity of @xmath0co does not indicate the amount of gas when gmcs start to pile up along the line of sight .
furthermore , because @xmath0co can be excited at a density as low as @xmath17 100 @xmath20 as a result of the photon trapping effect , the gas traced by @xmath0co is not necessarily associated with star formation .
finally , diffuse emissions that do not relate to star formation are involved in the tracers of sfr , such as uv , 24@xmath21 m , and h@xmath22 .
although @xcite show that the presence of diffuse emissions can alter the relation of @xmath5 and @xmath7 from linear to nonlinear , @xcite concluded that the contribution from the diffuse emissions alone is not sufficient to explain all the differences in the observed slopes .
inspired by the above issues , we study the k s law in a nearby galaxy ic 342 in terms of two aspects : the co - to - h@xmath1 conversion factors which depend on the metallicity and co intensity @xcite , and the utilization of an optically thin line @xmath3co ( 10 ) , which is reckoned to be a better tracer of the real mass distribution ( e.g. , @xcite ) .
ic 342 is a nearby face - on barred spiral galaxy located at 3.3 mpc .
1@xmath23 corresponds to 16 pc at that distance @xcite .
the co map made with a single dish shows a clear molecular bar with a length of approximately 4.5 kpc @xcite . a primary spiral arm emerging from the southern bar
is found in the co map .
ic 342 is undergoing bar - driven gas inflow @xcite which triggers the starburst at the center and therefore splits the physical environment from the galactic disk , providing a chance to study the k s law in diverse physical conditions of molecular gas .
the outline of this paper is as follows : the radio and infrared published data we used in this work are presented in section [ sec_data ] .
the study of the radial distribution of the gas and sfr is shown in section [ raddist ] .
the methods of deriving @xmath5 from @xmath0co ( 10 ) and @xmath3co ( 10 ) are demonstrated in section [ sec_xcolow ] and section [ ks13_high ] , respectively .
the uncertainties of the methods are discussed in section [ uncermass ] .
section [ ks ] presents the derived k s law with both @xmath0co ( 10 ) and @xmath3co ( 10 ) . in section [ galaxies_literatures ] ,
we examine the k - s law in other nearby galaxies using the published data of @xmath0co and metallicity to confirm the results from ic 342 .
the probable mechanisms of star formation in ic 342 are considered in section [ sf_ic342 ] .
the @xmath0co ( @xmath24 ) ( hereafter @xmath0co ) image of ic 342 ( figure [ fig_ico ] ) was taken from the nobeyama co atlas of nearby spiral galaxies @xcite .
the observations were made by the nobeyama 45-m telescope .
the angular resolution of the map is 20@xmath23 at 115 ghz , corresponding to a physical resolution of @xmath17320 pc .
the map extends up to a galactocentric radius of approximately 6.5 kpc ( 406@xmath23 ) .
the rms noise is approximately 1.5 k km s@xmath14 or @xmath17 5 m@xmath25 pc@xmath26 with the galactic standard co - to - h@xmath1 conversion factor ( @xmath4 @xmath10 2.3 @xmath11 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 ) @xcite .
the @xmath3co ( @xmath24 ) ( hereafter @xmath3co ) image ( yellow contours in figure [ fig_structure ] ) was also taken with the nobeyama 45-m telescope by @xcite .
the map covers the galactic center , the bar , and the spiral arm emerging from the southern bar end with a total area of @xmath17320@xmath23 @xmath11320 @xmath23 ( 5.1 kpc @xmath11 5.1 kpc ) , which is approximately one fourth of the @xmath0co map .
the angular resolution is also 20@xmath23 ( 320 pc ) at 110 ghz and the rms noise is 0.4 k km s@xmath14 .
the atomic hydrogen ( hi ) image of ic 342 is taken from @xcite .
the observations were carried out by the vla ( very large array ) using the c and d configurations .
the resulting angular resolution was 38@xmath23 @xmath11 37@xmath23 ( 610 pc @xmath11 590 pc ) at the wavelength of 21 cm .
the largest structure the observation sensitive to is about 15@xmath27 .
the amount of missing flux due to the short spacing problem was approximately 30% @xcite .
the hi image has a radius of 1300@xmath23 ( @xmath1721 kpc ) , which is far beyond the molecular disk of ic 342 by approximately a factor of three .
ic 342 is not included in the spitzer local volume legacy survey ( lvl ) and the spitzer infrared nearby galaxies survey ( sings ) , but in several individual programs , presumably because of its low latitude ( @xmath28=10@xmath2958 ) .
we used the 24@xmath21 m data from spitzer heritage archive ( sha ) as sfr tracer .
the data were obtained by the multiband imaging photometer ( mips ) @xcite .
the angular resolution of the 24 @xmath21 m map is @xmath175@xmath187 ( @xmath17 91 pc ) .
the hershel telescope provides infrared data from 70 to 500@xmath21 m .
the data are useful to derive the dust / gas temperature from the spectral energy distribution ( sed ) , which is required in our analysis ( [ ks13_high ] ) .
ic 342 is one of the samples in the project of key insights on nearby galaxies : a far - infrared survey with herschel ( kingfish ) @xcite .
the data are available on the project website .
herschel pacs and spire instruments provide the data at wavelengths of 70 , 100 , and 160 @xmath21 m ( pacs ) and 250 , 350 , and 500 @xmath21 m ( spire ) .
the original units of the images , the pixel sizes and the resolutions are listed in table [ ir_instrument ] .
the process of data reduction is described in the kingfish data products delivery users guide .
the neutral gas in a galaxy consists of molecular and atomic gas .
the former is required for star formation , but some nearby spiral galaxies harbor larger amounts of atomic gas relative to the molecular gas .
accordingly , we checked the content of the gas in ic 342 as the first step .
we use the image of @xmath0co to measure the radial distribution of @xmath5 . the inclination and
the position angle of the galaxy are 31@xmath30 and 37@xmath30 , respectively @xcite .
the step size between each azimuthal average data point is 10@xmath23 , starting from 0@xmath23 to 400@xmath23 ( 6.4 kpc ) or r / r@xmath31 @xmath32 0.30 , where r@xmath31 @xmath10 22.3@xmath27 or 21.4 kpc @xcite .
@xmath4 of 2.3 @xmath11
10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 @xcite is adopted to derive @xmath5 . the radial profile of @xmath5 is shown as crosses in figure [ fig_radialplot ] .
@xmath5 peaks at the galactic center ( @xmath17750 m@xmath25 pc@xmath26 ) and then gradually decreases to @xmath17 r / r@xmath31 = 0.05 .
a small bump at a radius of r / r@xmath33 ( @xmath172 kpc ) is caused by the brightest spiral arm . beyond the spiral arm
, @xmath5 decreases to @xmath1720 m@xmath25 pc@xmath26 at r / r@xmath31 = 0.3 .
the column density of atomic hydrogen ( hi ) is calculated by @xmath34 where @xmath35 is the brightness temperature @xcite .
the step size between each azimuthal average of hi data is 20@xmath23 ( 320 pc ) , starting from 0@xmath23 to 1300@xmath23 ( r / r@xmath31 @xmath32 0.97 )
. the radial profile of @xmath36 is shown in figure [ fig_radialplot ] with a plus symbol .
as seen in most spiral galaxies , the distribution of hi emissions extends well beyond the molecular disk .
@xmath36 hits the lowest point at the galactic center and shows an upward trend until r / r@xmath31 = 0.3 .
then , @xmath36 gently declines to r / r@xmath37 .
@xmath36 varies within a narrow range of 1 5 m@xmath25 pc@xmath26 , occupying @xmath17 35% of the total gas .
hence , ic 342 is an h@xmath1 dominant spiral galaxy , with potential for significant star formation .
the 24 @xmath21 m emissions trace the dust grains heated by the uv photons from young stars and therefore serve as a tracer of obscured sfr with a timescale no longer than 10 myr @xcite . on the other hand ,
h@xmath22 traces the ionized gas surrounding the massive stars and thereby is used as a tracer of unobscured sfr @xcite .
hence , 24 @xmath21 m + h@xmath22 is often used to obtain the total sfr in a given region ( e.g. , @xcite ) . however , because ic 342 is located behind the galactic plane ( @xmath28=10@xmath295 ) , h@xmath22 emissions suffer severely from dust extinction and , moreover , there is no available digital h@xmath22 image to access . for these reasons we use an infrared image as the sfr indicator .
the monochromatic infrared sfr is calculated as @xmath38^{0.885 } \label{equ_sfr}\ ] ] @xcite .
equation ( [ equ_sfr ] ) is calibrated for a local sfr with a spatial scale of @xmath17 500 pc and is valid in the range of @xmath39\lesssim 3\times 10^{44}$ ] .
ic 342 is within this range .
the uncertainty of the power and the coefficient are 0.02 and 15% , respectively .
we have smoothed the 24 @xmath21 m image from the resolution of @xmath176@xmath23 to 20@xmath23 to match with the co image .
the 24@xmath21 m image is also re - gridded to match with the co image for further analysis .
for this reason the step size and the radial range of @xmath7 ( triangles in figure [ fig_radialplot ] ) is the same as @xmath5 ( crosses ) .
the radial distribution of @xmath7 resembles that of @xmath5 in ic 342 .
the characteristic features of the radial distribution of @xmath5 , @xmath36 , and @xmath7 are summarized as follows : ( a ) the radial distribution of @xmath5 and @xmath36 reveal that ic 342 is a molecular - dominant galaxy ; ( b ) the molecular gas is concentrated in the center and the bright spiral arm at the radius of @xmath172 kpc ; ( c ) the radial distribution of @xmath7 resembles that of @xmath5 .
in this section , we examine the spatial distribution of molecular gas by using the galactic co - to - h@xmath1 conversion factor ( @xmath4 ) and a conversion factor ( @xmath2 ) that depends on the metallicity and line intensity @xcite .
first , the galactic @xmath4 of 2.3 @xmath11 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 was applied to the entire @xmath0co map .
the peak value of @xmath5 is 756 m@xmath25 pc@xmath26 at the galactic center .
such high @xmath5 is commonly found in the circumnuclear starburst @xcite .
the 3@xmath40 detection limit of @xmath0co corresponds to @xmath5 @xmath32 15 m@xmath25 pc@xmath26 .
the mean @xmath5 of ic 342 is 36 m@xmath25 pc@xmath26 , principally contributed by the galactic disk .
both observations ( e.g. , @xcite ) and simulations ( e.g. , @xcite ) suggest that metallicity is a key parameter affecting @xmath4 . in metal - poor regions
, @xmath4 tends to be higher than the galactic @xmath4 because of the photodissociation of co. furthermore , @xcite show that @xmath4 depends on the co intensity as well .
such dependence of @xmath4 on co intensity is also found within a single galaxy @xcite .
therefore , we adopt the @xmath4@xmath41@xmath42 relation derived by @xcite : @xmath43 where @xmath44 is the @xmath0co line intensity measured in k km s@xmath14 and @xmath41 is the metallicity divided by the solar metallicity ( @xmath45 ) .
the conversion between @xmath41 and the commonly used form of metallicity of @xmath46 $ ] is @xmath47 -8.76 , \label{equ_zgrad}\ ] ] where the solar metallicity on a @xmath48 scale is assumed to be 8.76 @xcite .
equation ( [ equ_xco ] ) is available from sub - kpc scale to unresolved galaxies @xcite . for ic 342 , as in most of galaxies ,
the gradient of metallicity is measured from the hii regions in various radii of the galaxy .
@xcite derived the oxygen abundance for the hii regions observed by @xcite and fitted the gradient with the form : @xmath49 = 12+\log [ \mathrm{o / h}]_{0 } + c_{\mathrm{o / h}}\times ( r / r_{25 } ) , \label{equ_ohgrad}\ ] ] where @xmath46_{0}$ ] is the extrapolated central oxygen abundance , and @xmath50 is the slope of the oxygen abundance gradient with respect to the fractional radius @xmath51 .
the hii regions in ic 342 indicate a clear negative gradient of the abundance with increasing radius .
the value of @xmath46_{0}$ ] is 8.85 , and @xmath50 is 0.9 .
the uncertainty of the derived value of @xmath52 $ ] is 0.12 .
the metallicity of each individual pixel is calculated and substituted into equation ( [ equ_xco ] ) and ( [ equ_zgrad ] ) to obtain @xmath2 at each position .
@xmath2 has a range ( 1 6 ) @xmath11 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 . figure [ fig_xcoxcov ] displays a correlation of @xmath5(@xmath4 ) versus the ratio of @xmath5(@xmath2)/@xmath5 ( @xmath4 ) or @xmath2 .
the figure indicates that @xmath2 @xmath53 @xmath4 when @xmath5(@xmath4 ) @xmath19 100 m@xmath25 pc@xmath26 , and vice versa .
in addition , @xmath5(@xmath4 ) @xmath32 @xmath5(@xmath2 ) occurs at @xmath17100 m@xmath25 pc@xmath26 .
the @xmath2 map is shown in figure [ fig_xco ] . in the galactic center
, @xmath2 is as low as 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 , consistent with the value measured by the high resolution c@xmath54o , 1.3 mm dust continuum , and the virial theorem in ic 342 @xcite .
thus the utilization of the galactic standard @xmath4 leads to the overestimation of @xmath5 at the central region .
the peak @xmath5 becomes 344 m@xmath25 pc@xmath26 by using @xmath2 , which is approximately two times lower than the mass obtained from the galactic @xmath4 .
the compact regions in the spiral arms have @xmath2 close to the galactic @xmath4 .
the remaining diffuse parts have @xmath2 @xmath19 3 @xmath11 10@xmath12 @xmath13 ( k km s@xmath14)@xmath14 .
eventually , the detection limit ( 3 @xmath40 detection ) of @xmath5 based on @xmath2 is about 26 m@xmath25 pc@xmath26 and the mean @xmath5 is approximately 51 m@xmath25 pc@xmath26 .
the radial distribution of @xmath5 based on @xmath2 is plotted as circles in figure [ fig_radialplot ] .
@xmath36 is still negligible in the central region and the disk .
the velocity - integrated intensity map of @xmath3co acquired from @xcite is shown in the left panel of figure [ fig_structure ] with yellow contours .
the @xmath3co to @xmath0co ratio is 0.05 0.2 @xcite .
owing to the low intensity of @xmath3co , only the pixels with significant detection are considered . the central @xmath171@xmath27 corresponding to the galactic nucleus found in the high - resolution map @xcite is defined as _
galactic center_. this region has at least 15 @xmath40 detections .
the central region is colored purple in the right panel of figure [ fig_structure ] . apart from the galactic center , the pixels with @xmath55 5 @xmath40 detection are taken as well .
the criterion chiefly captures the primary spiral arm and the bar end .
we refer to this region as the _ spiral arm_. this region is colored orange in figure [ fig_structure ] .
provided that @xmath3co is in local thermal equilibrium ( lte ) and optically thin , the column density of h@xmath1 is computed by @xmath56}{[^{13}co]}}\times \frac{i_{\mathrm{^{13}co}}}{1-e^{-5.29/t\mathrm{_{ex } } } } , \label{equ_n13co } \end{aligned}\ ] ] where @xmath57 is the optical depth of the @xmath3co , [ h@xmath1]/[@xmath3co ] is the inverse of the @xmath3co abundance , @xmath59 is the intensity of @xmath3co in k km s@xmath14 , and @xmath60 is the excitation temperature in kelvin @xcite .
as shown in equation ( [ equ_n13co ] ) , prior knowledge of the optical depth , @xmath3co abundance , and gas ( excitation ) temperature are required .
the optical depth is estimated by @xmath61 in which @xmath0co is assumed to be optically thick and @xmath3co is optically thin @xcite .
the isotopic abundance ratio [ @xmath0c]/[@xmath3c ] is assumed to be @xmath32 30 - 50 from the measurements for the galactic center of ic 342 @xcite . given that [ @xmath0c]/[@xmath3c ] @xmath10 [ @xmath0co]/[@xmath3co ] ( e.g. , @xcite ) and [ @xmath0co]/[@xmath62 is approximately ( 5 8) @xmath11 10@xmath63 , we adopt a constant [ @xmath3co]/[@xmath62 of 1.5 @xmath11 10@xmath64 in all regions concerned .
the value is consistent with the mean @xmath3co abundance observed in the galactic star forming clouds @xcite .
the gas temperature is also required to obtain the column density of h@xmath1 .
the assumption that _ the gas temperature ( @xmath65 ) is equal to the dust temperature ( @xmath66 ) _ is made .
then , the dust / gas temperature can be yielded by the infrared sed .
the sed is constructed with the 24@xmath21 m observations by spitzer ( [ sec_data_sp ] ) and the 70
500@xmath21 m observations by herschel ( [ sec_data_her ] ) .
all images are converted to janskys and then convolved to the lowest resolution among the images , i.e. , fwhm = 37@xmath23 of the herschel 500 @xmath21 m map .
the method of convolution kernels between spitzer and herschel images is described in @xcite .
we used the kernels and the idl procedure @xmath67 @xcite to convolve the images .
then , a two - component planck distribution @xmath68+c_{\mathrm{c}}\ ; \left [ \lambda ^{-\beta_{\mathrm{c}}}\ ; b_{\nu}\left ( t_{\mathrm{c } } \right ) \right ] \label{equ_2compt}\ ] ] was assumed as the shape of sed , where @xmath69 is a planck function , @xmath70 and @xmath71 are the temperatures of the warm and cold components , respectively , and @xmath72 and @xmath73 are the scaling constants of each component @xcite .
moreover , the planck functions were modified by the dust emissivity index @xmath74 . for the warm component , the warm dust emissivity index ( @xmath75 )
was fixed at two as suggested by @xcite . on the other hand , the cold dust emissivity index ( @xmath76 )
is a free parameters to be fit .
@xmath76 is in the range of 1 3 , so the program is forced to find the best value of @xmath76 in this range . to summarize
, there are five free parameters to fit : @xmath72 , @xmath73 , @xmath70 , @xmath71 , and @xmath76 .
finally , the idl function @xmath77 @xcite is used to carry out a levenberg
marquardt least squares fit of the data .
@xmath77 requires a user - supplied function for fitting . in this case ,
equation ( [ equ_2compt ] ) is the function employed . since the images have a final resolution significantly lower than our co maps , seds for the global ( center + arm ) , galactic center and arm are obtained with the total flux in each defined region in figure [ fig_structure ] , instead of a pixel - by - pixel fitting .
the seds are displayed in figure [ fig_t ] .
the derived @xmath76 is in the typical range of 2.0 2.4 .
the galactic center has a higher temperature in both warm and cold components ( warm : 64.7 @xmath78 1.6 k , cold : 25.1 @xmath78 0.4 k ) than those of the spiral arm ( warm : 56.3 @xmath78 1.1 k , cold : 19.0 @xmath78 0.5 k ) .
the global sed represents the average of the center and disk with @xmath70 @xmath10 62.1 @xmath78 1.3 k and @xmath71 @xmath10 23.7 @xmath78 0.8 k. @xcite derived @xmath79 and @xmath80 in eleven nearby galaxies with similar methods and similar datasets .
the temperatures of ic 342 are compatible with other galaxies .
the cold dust component is associated with the dust , molecular , and atomic hydrogen clouds mostly heated by the interstellar radiation field ( isrf ) . on the other hand
, the warm dust component is heated by the ob stars in hii regions @xcite .
accordingly , @xmath71 ( @xmath1725 k for the center and @xmath1719 k for the arm ) is substituted into equation ( [ equ_n13co ] ) to derive @xmath5 based on @xmath3co .
the resulting @xmath5 lies within 30 400 @xmath81 pc@xmath26 .
the mean @xmath5 is 90 @xmath81 pc@xmath26 .
the mean @xmath5 obtained by @xmath3co is significantly larger than those from @xmath0co ( [ sec_xcolow ] ) because the observations and the selection of available pixels are biased to the co bright region as a result of the weak intensity of @xmath3co .
@xmath5 obtained from @xmath0co(@xmath2 ) and @xmath3co are compared in figure [ fig_massmass ] .
only the corresponding pixels in the @xmath0co map with significant @xmath3co detection ( figure [ fig_structure ] right panel ) are selected for comparison .
@xmath5 derived from both lines are consistent where @xmath5 @xmath19 100 m@xmath25 pc@xmath26 .
such high @xmath5 is dominated by the central region of the galaxy .
however , @xmath5 acquired from @xmath3co is smaller than that from @xmath0co at @xmath5 @xmath82 100 m@xmath25 pc@xmath26 .
the deviation of @xmath5 between two lines can be attributed to the overestimation of the @xmath3co abundance by adopting the central value even outside the center .
the uncertainty of the @xmath3co abundance will be discussed in the next section ( [ uncertainty_cabundance ] ) .
nevertheless , we should note that in the spiral arm , @xmath0co and @xmath3co peaks do not coincide with each other ( see figure [ fig_structure ] ) .
therefore , it is not necessary to expect a consistent result from these lines .
to obtain the @xmath5 map based on @xmath0co , the gradient of oxygen abundance is required .
the emission lines of hii regions spanning a large fraction of the galactic disk are needed to construct the metallicity gradient of a galaxy .
the relevant studies have demonstrated that the metallicity ( 12+[o / h ] ) is a function of the galactocentric radius in spiral galaxies , although the slopes can vary between positive and negative @xcite . for ic 342 ,
only five hii regions were observed to infer the metallicity gradient @xcite .
it seems that the gradient is affected by the number of hii regions observed .
@xcite , however , derive the gradient of metallicity with eight hii regions for ngc 6946 , classified as sab(rs)cd as ic 342 , and the result is consistent with the results derived with more than 200 hii regions in the galaxy @xcite .
the comparison demonstrates that the variation of oxygen abundance is on a global scale and the local fluctuation is not significant .
the uncertainty of @xmath2 attributed to the oxygen abundance alone is @xmath1740% .
the uncertainty is dominated by the scatter of the measured oxygen abundance from the general trend at each position .
the isotopic ratio [ @xmath0c]/[@xmath3c ] grows with radius in our galaxy @xcite .
however , a constant @xmath3co abundance measured for the center is adopted for the entire region in this work because of the lack of abundance ratio along the galactocentric radius of ic 342 .
@xmath3c is produced during the cno cycles of stars . in an active star forming region , e.g. , galactic center , @xmath3c is produced faster , influencing the interstellar isotope abundances .
there are two subsequent mechanisms determining the @xmath3co abundance : fractionation reaction and photodissociation .
the fractionation reaction @xmath83 enriches @xmath3co but requires relatively low temperature ( @xmath53 35k ) @xcite and uv photons to provide the ionized carbons , preferentially happening near the uv - exposed edges of gmcs .
however , uv photons cause the photodissociation of both @xmath0co and @xmath3co as well ( @xmath84 ) . because self - shielding is more effective for h@xmath1 than co lines , photodissociation can reduce the @xmath0co and @xmath3co abundance relative to h@xmath1 .
in addition , because @xmath0co has more effective self - shielding and larger abundance , the photodissociation destroys @xmath3co faster than @xmath0co . combining both phenomena
, the abundance of @xmath3co depends on the relative rate of the fractionation and photodissociation.the combined effect can be small , and these phenomena compensate for one another @xcite , or leads to the underestimation of @xmath85 by 30%50% @xcite .
providing that [ @xmath0c]/[@xmath3c ] ( or [ @xmath0co]/[@xmath3co ] as we have assumed ) has the same gradient as our galaxy @xcite , the isotopic ratio then increases from @xmath1740 at the galactic center to @xmath1750 at the spiral arm at 2 kpc .
this deviation in the isotopic ratio is smaller than the uncertainties suggested by the above chemical processes .
we have insufficient information to quantify the actual error .
a maximum error of 50% contributed from the uncertainty of @xmath3co abundance is quoted in this work , but such large uncertainty is unlikely to be true if the abundance gradient is similar to that of the milky way .
the interaction and energetics of gas and dust are complex . in terms of simulations
, @xcite made starburst models to correlate the observed fir / submm / mm properties of gas and the star formation history of the starburst galaxy .
the model predicts that @xmath65 is close to @xmath66 in high - column - density regions .
@xcite use lvgs model to calculate the radiative transfer of the cooling and heating of gas and dust .
the result shows that in the high - volume - density region with @xmath86 10@xmath87 @xmath20 the gas and dust grains become well coupled and @xmath65 @xmath32 @xmath66 .
recent work has also suggested a similar threshold of 10@xmath88 @xmath20@xcite for good coupling between @xmath65 and @xmath66 . even though the critical density of @xmath3co is @xmath1710@xmath89 @xmath20 ,
tracing the gas with density @xmath19 10@xmath89 @xmath20 , @xmath3co is useful to locate the dense cores embedded in gmcs @xcite , which have a typical density of @xmath1710@xmath90 @xmath20 ( e.g. , @xcite ) .
therefore , it is not surprising to see that @xmath3co coexists with 24 @xmath21 m emissions in both the spiral arm and the center of ic 342 @xcite .
hence , the assumption of @xmath65 @xmath32 @xmath66 is rather fair for the molecular gas traced by @xmath3co in terms of gas density . the inferred dust / gas temperature from sed fitting is @xmath17 20 25 k in the infrared bright galaxy ic 342 .
although the temperature is consistent with the temperature of the cloud complexes in other infrared bright galaxies ( e.g. , @xcite ) , the galactic gmcs have typical temperatures of @xmath17 10 k , which is approximately two times lower than that of these infrared bright galaxies . nonetheless , it is unclear whether the external galaxies , especially the infrared bright galaxies , hold gas properties similar to those in our milky way ( e.g. , @xcite ) .
because @xmath0co is detected throughout the entire galaxy ( figure [ fig_ico ] ) , the map is first used to perform the k s plot by using the value of @xmath5 based on @xmath4 and @xmath2 .
the fitting of the k s plot is dominated by the area with @xmath5 @xmath91 100 m@xmath25 pc@xmath26 because of the bulk of data points in this range .
we therefore refer to the results as _ the k s law in the low-@xmath5 region_. on the contrary , considering the fact that @xmath3co was only observed toward the co bright regions , the k
s plots made with @xmath3co and the corresponding @xmath0co pixels are referred to as _ the k
s law in the high-@xmath5 region_. since the fitting procedures can alter @xmath6 of k s law ( @xcite and references therein ) , the routines we employed are described below .
the @xmath92 routine @xmath93 @xcite is used to implement all the fitting of the k - s plots in this work . similar as the @xmath77 in [ ks13_high ] , @xmath93 allows a user - defined function , then performs levenberg - marquardt least - squares minimization for the dataset .
we fitted the data with a single power law , i.e. , @xmath94 , where @xmath95 , @xmath96 , @xmath6 is the slope of the k s plot , and
@xmath97 is a constant . because both @xmath5 and @xmath7 contain measured uncertainties , a model function @xmath98 from the @xmath93 library is utilized .
@xmath98 is developed to handle a linear fitting to the data with uncertainties in both directions , following the methodology of _ numerical recipes_. the k
s law is examined with @xmath5 derived with the galactic standard @xmath4 in [ mass12const ] and the sfr from equation ( [ equ_sfr ] ) ( the 24 @xmath21 m image has been re - gridded to the same spacing and the resolution as the @xmath0co map in [ subsec_sfr ] ) . the k
s plot is shown in the left panel of figure [ fig_largespatial_ks ] .
gray , red , orange , yellow , and green indicate the contours of 1 , 10 , 40 , 65 , and 100 data points per 0.1 dex - wide cell of both @xmath5 and @xmath7 , respectively .
fifty percent of the data points are inside the yellow area with @xmath5 @xmath32 15 ( sensitivity limit ) 40 m@xmath25 pc@xmath26 .
thus the result is controlled by the low @xmath5 region .
the slope of the k
s plot ( @xmath6 ) is 1.11 @xmath78 0.02 .
@xcite created k s plots for seven nearby spiral galaxies by the same method but with @xmath0co ( 21 ) , assuming a constant ratio of @xmath0co ( 21)/(10 ) .
@xmath5 in their samples mostly lie in the range of 350 m@xmath25 pc@xmath26 .
the lower end of @xmath5 in @xcite is smaller than our estimate as a result of better sensitivity .
the samples in @xcite have a physical resolution of 750 pc , approximately two times larger than in this study .
previous works show that the physical resolution can affect @xmath6 of the k s law owing to the evolution of gmcs . for a small scale ( @xmath17 60 100 pc ) ,
discrete star forming systems have different evolutionary stages .
for example , the youngest gmcs have no star formation but large amounts of molecular gas ( large @xmath5 , low @xmath7 ) , whereas old gmcs have the opposite characteristics ( small @xmath5 , large @xmath7 ) .
these populations then scatter the k s relation @xcite .
our resolution of 320 pc and the 750 pc of @xcite are considerably larger than the physical sizes of the gmcs ; therefore , the abovementioned situation does not exist .
instead , the k s law represents a time - averaged relation , averaging many elements with various evolutionary stages , and thus the scaling relation of @xmath5 and @xmath7 can be maintained @xcite .
@xmath6 of 1.11 @xmath78 0.02 in ic 342 is comparable to the average slope of 1.0 @xmath78 0.2 among the galaxies in @xcite .
@xmath6 of unity implies that the gas depletion time ( @xmath99 ) and star formation efficiency ( sfe ) are independent of @xmath5 .
the @xmath99 of ic 342 is about 2 @xmath11 10@xmath100 yr , or sfe @xmath17 5 @xmath11 10@xmath101 yr@xmath14 ( figure [ fig_largespatial_ks ] left panel ) .
in conclusion , on the basis of the comparison with the larger sample in @xcite , we suggest that the relation of the gas and star formation properties in ic 342 are not considerably different from those of other galaxies . however , we should note that in addition to the influence from the intrinsic properties of gmcs @xcite , @xcite suggest that the configuration of gmcs and stars can affect @xmath6 of star formation law when observations are made with approximately sub - kiloparsec resolution . under this physical resolution ,
the distribution of molecular gas is spatially resolved , but individual gmcs are not . as a result ,
the observed @xmath6 is a convolution of the intrinsic star formation mode , cloud mass spectrum in the sampled area , volume filling factor , and region size ( physical resolution ) @xcite .
these parameters may vary among individual galaxies .
in general , @xmath6 will decrease when the physical resolution increases , although we did not see the significant change among ic 342 ( resolution of 320 pc ) and the samples of @xcite ( resolution of 750 pc ) .
we created a k
s plot using the same method as in [ largeconstant ] but with @xmath2 instead .
the result is displayed in the right panel of figure [ fig_largespatial_ks ] .
the color scale is the same as in [ largeconstant ]
. fifty percent of the data points are located inside the green region ranging from 25 60 m@xmath25 pc@xmath26 .
the figure shows that the relation between @xmath5 and @xmath7 is steepened .
the derived @xmath6 is 1.39 @xmath780.03 .
because of the superlinear relation , the @xmath99 and the sfe are no longer constant
. specifically , @xmath99 declines with increasing @xmath5 , namely , the sfe rises when @xmath5 increases . because the fitting is dominated by the bulk of the data points at low-@xmath5 regions ( around 25 60 m@xmath25 pc@xmath26 ) , the high - surface - density regions with @xmath5 @xmath19 100 m@xmath25 pc@xmath26 are located beyond the best fit , suggesting a steeper trend . in this section
we study the k s law in the area with high @xmath5 .
@xmath3co map can be used as a filter to select the region with high @xmath5 because ( 1 ) @xmath3co is a weak line , the location where @xmath3co can be significantly detected is biased to high-@xmath5 regions , and ( 2 ) owing to its higher critical density , @xmath3co is one of the transitions available for locating the dense cores in gmcs .
these dense cores have higher column density ( @xmath5 ) and volume density prepared for future star formation . in the right panel of figure [ fig_structure ] , the pixels with @xmath3co detection greater than 5@xmath40 are masked with color .
these pixels are used to derive a k s plot based on @xmath3co . with the surface density derived in [ ks13_high ] , a k
s plot in terms of @xmath3co is displayed in figure [ fig_ks13 ] .
in this domain , @xmath5 ranges from approximately 30 400 m@xmath25 pc@xmath26 .
@xmath99 ( or sfe ) changes by more than one order magnitude over this range , from @xmath17 1.5 @xmath11 10@xmath100 yr at @xmath5 @xmath32 30 m@xmath25 pc@xmath26 to @xmath1710@xmath102 yr at @xmath5 @xmath32 400 m@xmath25 pc@xmath26 .
@xmath6 of the k
s plot is derived as 1.78 @xmath78 0.07 .
the best fit of the low-@xmath5 regions based on @xmath0co ( @xmath2 ) is over - plotted in figure [ fig_ks13 ] .
it can be seen that the @xmath7 of high-@xmath5 regions lie beyond the best fit of low-@xmath5 regions , implying a higher sfr at a given @xmath5 .
the corresponding pixels in the @xmath0co map with significant @xmath3co emissions are used to create the k
s plot at high-@xmath5 region for the comparison .
first , the k s plot is created with the galactic @xmath4 .
the result is displayed in the left panel of figure [ fig_ks12 ] .
@xmath5 lies between 20 to 800 m@xmath25 pc@xmath26 .
@xmath6 is 1.33 @xmath78 0.02 with all data , slightly larger than the slope in the low-@xmath5 region ( @xmath0co , @xmath4 ) .
the k s plot with @xmath2 is shown in the right panel of figure [ fig_ks12 ] .
the derived @xmath6 is approaching 2 ( 1.94 @xmath78 0.11 ) with all data points .
@xmath6 derived in this section are larger than those in the environment of low-@xmath5 , regardless of whether @xmath4 or @xmath2 is used .
the variation of the power law index of the k s law in this section is summarized in table [ tab_ksslpoes ] .
there are two conclusions drawn from the above k - s plots .
first of all , the adoption of @xmath2 steepens the k s plots in both low- and high-@xmath5 regimes .
this is reasonable to expect from the difference in @xmath4 and @xmath2 ( figure [ fig_xcoxcov ] ) .
secondly , there is a trend that @xmath6 would increase with increasing @xmath5 .
the steeper slope in the higher-@xmath5 area is confirmed by the independent measurement of @xmath3co .
a quantitative analysis of the dependence of @xmath6 on the @xmath5 considered is displayed in figure [ fig_n_bound ] .
figure [ fig_n_bound ] displays the derived @xmath6 with a different lower threshold of the used data .
only the data greater than the defined threshold of @xmath5 are used .
in other words , the dynamic range of fitting is narrowed down from _ all data _ to _
@xmath103 @xmath19 130 m@xmath25 pc@xmath26 _ , with intervals of @xmath19 60 , 100 , and 130 m@xmath25 pc@xmath26 .
the analysis was performed on three datasets , @xmath0co ( @xmath4 ) , @xmath0co ( @xmath2 ) , and @xmath3co . even though @xmath6 with higher thresholds suffer from large uncertainties because of the scatter of measurements and limited data points ,
there is a clear trend that the slopes rise with an increasing threshold . in the low-@xmath5 region ( squares ) , @xmath6 is generally less than two . in the environment of @xmath5 @xmath19 100 m@xmath25 pc@xmath26 , @xmath6 increases to around two to three .
such a phenomenon is observed in both transitions .
various studies have concluded that the sfr is well correlated with the amount of dense gas .
the most straightforward evidence is the linear correlation between @xmath104 and @xmath105 ( or @xmath106@xmath7 ) .
the correlation is held from galactic dense cores to extragalactic unresolved sources @xcite .
simulations of turbulent molecular clouds show that the sfr is controlled by the density probability distribution function ( pdf ) ( e.g. , @xcite ) .
the galactic turbulence can be driven by gravitational instability , shocks or interaction of galaxies .
@xcite simulates the pdf of different gas surface densities across @xmath17 20 2000 m@xmath25 pc@xmath26 .
the result shows that the fraction of dense gas ( i.e. , sufficiently dense for star formation ) increases with gas surface density ( see figure 3 of @xcite ) .
such a trend shown by the simulations has already been observed in the molecular clouds of our milky way ( e.g. , @xcite ) .
hence , the larger @xmath6 in the higher-@xmath103 regime indicated by low density tracers ( e.g. , co ) may be a reflection of the increasing dense gas fraction .
the results from the study of ic 342 show that the k s law may not be identical within a galaxy , namely , the slope steepens up toward the high-@xmath85 region , reflecting a shorter @xmath107 ( or higher sfe ) and perhaps a different star formation mechanism .
we examine whether the results of ic 342 are particular to this galaxy or a general situation by using the published data of some nearby galaxies .
there are two selection criteria for sample galaxies . at first
, the galaxy has to be studied in terms of the metallicity based on the framework of the p - method @xcite .
the candidates reaching this criterion are shown in @xcite and @xcite .
@xcite and @xcite derive the metallicity gradient with hii regions for their sample galaxies .
@xcite also derive the metallicity of the nuclear and circumnuclear regions for some of their samples .
secondly , the galaxy must be studied for @xmath85 and @xmath108 . @xmath85 and
@xmath108 were compiled from literature and divided into two groups .
the first group includes ten galaxies with published radial @xmath109 , @xmath85 , and @xmath108 from @xcite.the physical resolution of the observations is @xmath17800 pc . @xmath85 and @xmath108 with @xmath193@xmath40 detection
furthermore , for a fair comparison with ic 342 , only the radii with @xmath85@xmath19@xmath109 are used , i.e. , the h@xmath1-dominated regions . with the above two criteria ,
some data points at the outskirts of the galaxies are eliminated .
the galaxies ( ngc 628 , ngc 3184 , ngc 3198 , ngc 3351 , ngc 3521 , ngc 4736 , ngc 5055 , ngc 5194 , ngc 6946 , and ngc 7331 ) in the first group will be used as the samples of _ galactic disks _ with lower @xmath85 .
the second group comprises nine _ starburst galaxies _ from @xcite , representing the high-@xmath85 domain .
most of starburst samples have co and sfr observations with sub - kpc resolution towards their circumnuclear disks . in such active star forming regions
, it is reasonable to assume @xmath85 @xmath110 @xmath109 , as observed in the center of ic 342 .
the metallicity of the starburst samples are from either the circumnuclear metallicity in @xcite , or the extrapolation of the gradient toward the center from @xcite .
the second group contains ngc 253 , ngc 1097 , ngc 2093 , ngc 3034 , ngc 3351 , ngc 4736 , ngc 5194 , ngc 5236 , and ngc 6946 .
s laws are examined on the basis of the galactic standard @xmath4 and @xmath2 calculated from the @xmath0co intensity and the metallicity .
the results are shown in figure [ fig_manygalaxies ] .
@xmath6 for _ disk _ , low-@xmath85 samples with @xmath4 ( green circles ) is 0.96 @xmath78 0.06 and 1.35 @xmath78 0.07 for the same samples but using @xmath2 ( red circles ) .
both values are consistent with the slopes of ic 342 in the low-@xmath85 regime ( 1.11 @xmath78 0.02 and 1.39 @xmath78 0.03 , respectively ) . for the circumnuclear starbursts , @xmath85 is greater than 100 m@xmath25 pc@xmath26 , whether @xmath4 or @xmath2 is used .
@xmath6 is 1.88 @xmath78 0.14 and 2.32 @xmath78 0.13 by using @xmath4 and @xmath2 , respectively .
both numbers are in good agreement with the value of @xmath6 derived from ic 342 with a threshold of 100 m@xmath25 pc@xmath26 , i.e. , @xmath6 = 1.96 @xmath78 0.16 and 2.65 @xmath78 0.22 for @xmath85 ( @xmath0co , @xmath4 ) and @xmath85 ( @xmath0co , @xmath2 ) , respectively .
the steeper @xmath85@xmath108 relation toward high-@xmath85 regime is not only found in local circumnuclear regions as this work shows , but also in more extreme objects ( e.g. , ulirgs ; @xcite ) and high redshift galaxies ( e.g. , @xcite ) . by comparing disk - averaged k - s law between ulirgs and normal disk galaxies
, @xcite also suggest that k - s law is not uniform among different star forming populations .
because of their high sfe , @xmath6 is larger in ulirgs than in normal disk galaxies
. thus , we then have a certain level of confidence that the results of ic 342 are not particular but universal for at least nearby spiral galaxies . the slopes in this section
are summarized in table [ tab_ksslpoes ] as well .
the slope of the k
s plot depends on the mechanisms of star formation as mentioned above .
we discuss two possible mechanisms of star formation in ic 342 , gravitational instability and cloud - cloud collisions , as well as the theoretical aspects of changing the power law index of k s law as a result of the properties of gmcs . for star formation caused by gravitational instability
, it is assumed that a constant fraction of molecular gas will be converted to stars per free fall time . because @xmath111 and @xmath112 , where @xmath113 and @xmath114 are gas mass and sfr per unit volume , @xmath114 is proportional to @xmath115 , or @xmath116 , assuming that the scale height of the gas disk is constant in galaxies .
this naturally explains the well - known `` disk - averaged '' k s law with a power law index of 1.4 in @xcite .
our result of the k
s plot of 1.39 @xmath78 0.03 in the low-@xmath5 region ( [ kslowsig ] ) with @xmath2 is consistent with the slope expected by the theory and simulations of the collapse of a molecular cloud through gravitational instability @xcite .
we then investigate the relation between the gravitational instability of gas and star formation in ic 342 via the toomre q parameter @xcite and 24@xmath21 m image .
@xcite suggests that the gas can collapse when @xmath117 exceeds the critical density ( @xmath118 ) , leading to a toomre - q parameter less than unity because @xmath119 where @xmath117 is @xmath120 .
@xmath118 is determined by the epicyclic frequency @xmath121 and velocity dispersion @xmath40 as @xmath122 an hi map of ic 342 was observed by @xcite using the vla ( see [ sec_data_hi ] ) .
the rotation curve was derived with the brandt model in @xcite with the hi data .
the maximum velocity of @xmath17170 km s@xmath14 is located at the radius of @xmath175 kpc .
@xmath121 at a radius of @xmath123 is calculated by @xmath124 .
the map of hi velocity dispersion @xcite is used to estimate @xmath40 in equation ( [ sigcrit ] ) .
@xmath40 has a range approximately 7 25 km s@xmath14 . to fairly derive the total gas mass at each pixel , the @xmath0co ( @xmath2 ) and 24@xmath21 m images were convolved to the same resolution as the hi image ( 38@xmath23 @xmath11 37@xmath23 )
firstly , the radial distribution of the q parameter is displayed as a solid black line in figure [ raddist ] .
the radial profile shows that only at the galactic center q is @xmath53 1 . outside the galactic center ,
the average q parameter varies in a range of 1 2 .
because the gas distribution of ic 342 is not symmetric at each position angle of the galaxy , the radial distribution may smooth the real distribution of the q parameter .
for this reason , we then conduct the spatial distribution of the q parameter .
figure [ fig_toomreq ] shows the spatial distribution of the toomre q parameter in a color scale overlaid on the 24@xmath21 m image in contours .
the regions prone to collapse ( q @xmath53 1 ) are shown in color . in the innermost region of the galaxy , the bar end , and several knots in the spiral arms , q @xmath32 0.5 - 0.8 .
the region where q @xmath53 1 is in agreement with the peaks of 24@xmath21 m , indicating that the stars in the disk are likely formed through gravitational instability .
indeed , @xcite suggest that @xmath1960% of massive young stellar objects are inhabiting in the regions where the gas is unstable .
star formation is known to potentially occur in regions with efficient cloud - cloud collisions @xcite .
the collision time scale is @xmath125 , where @xmath126 is the number density of clouds , @xmath127 is the relative velocity of the clouds , and @xmath40 is the average cloud cross section .
@xmath128 per collision time is then proportional to @xmath129 .
@xcite suggest that there are @xmath17 100 molecular cores in the central region of ic 342 and the cores are colliding with each other . the origin of the collisions can be attributed to bar - induced orbital crowding .
the shocks produced after the cloud - cloud collisions have been found using shock tracer molecules , e.g. , ch@xmath130oh and hnco @xcite .
these observations highlight the region where cloud - cloud collisions are promoted : along the leading side of the bar and the intersections of the bar and the circumnuclear ring ( these two regions are located in the region defined as the center in figure [ fig_structure ] , with a diameter of @xmath17 1@xmath27 ; @xcite ) .
indeed , both regions are associated with dense gas ( e.g. , hcn , @xcite ) and star forming tracers , such as h@xmath22 , and ir / radio star forming regions ( e.g. , @xcite ) . even though star formation by cloud - cloud collisions is suggested by previous observations with various lines ,
the possibility of gravitational instability can not be ruled out because the tommre q parameter is well below unity in the central region of the galaxy ( figure [ fig_toomreq]).the two mechanisms may work together @xcite .
that is , the gravitationally bound molecular clouds ( @xmath1710@xmath131 m@xmath25 ) are formed through the gravitational instability ; at the same time , cloud - cloud collisions are taking place between the clouds to form giant molecular cloud associations ( gmas ) .
this scenario is able to explain the high sfe of the spiral arm in m51 @xcite .
if the above scenario take places in the high-@xmath132 region of ic 342 , @xmath6 = 2 3 in this region represents a mix of the mechanisms .
in addition to the difference in the effective star formation mechanism , the k s law may be altered because of the change in the intrinsic properties of gmcs .
@xcite have performed theoretical calculations of a two - component star formation law regardless of the galactic - scale process .
the transition of @xmath132 is approximately 85 m@xmath25 pc@xmath26 . in the regions where the @xmath132 is greater than @xmath1785 m@xmath25 pc@xmath26 ,
the galactic ism pressure is sufficiently large to be comparable with the internal pressure of gmcs , and thus the density of gmcs is forced to increase with growing @xmath132 to balance the pressures . at the same time , the free - fall time of the gmcs is reduced @xcite , resulting in faster star formation . on the other hand , at low-@xmath132 regions ,
gmcs are independent of the environment so the @xmath128 is determined only by the gmcs themselves , implying a flattened k s relation .
we studied the kennicutt - schmidt ( k s ) law in the nearby barred spiral galaxy ic 342 with published @xmath0co ( 10 ) , @xmath3co ( 10 ) , and infrared data .
the main results are summarized as follows : 1 . after correcting for the oxygen abundance and @xmath0co intensity , the @xmath0co - to - h@xmath1 conversion factor ( @xmath2 )
was found to be 23 times lower than the galactic standard @xmath4 in the center of ic 342 and @xmath172 times higher than the @xmath4 in the galactic disk ( [ sec_xcolow ] ) .
2 . the power law index of the k
s plot is approximately 1.4 in the low-@xmath85 ( @xmath0co,@xmath2 ) regions ( often 25 60 m@xmath25 pc@xmath26 ) , and increases to approximately 2 3 at high-@xmath85 ( @xmath0co,@xmath2 ) regions ( around @xmath19 100 m@xmath25 pc@xmath26 ) ( [ sec_largespatialksxvar ] , [ ks12var ] ) .
the larger slope in the environment of high surface density is confirmed with independent measurement of @xmath3co ( [ secks13 ] ) .
3 . by setting different lower limits of the dynamic range for fitting
, we confirmed that the slope of the k
s law is gradually steepened with increasing @xmath85 , presumably as a result of the increase in the dense gas fraction ( [ sec_environ_var ] ) .
4 . using the published data of @xmath85 , @xmath108 ( measured at the sub - kpc scale ) , and the metallicity of nearby galactic disks and starburst centers , we confirm that the environmental variation in the k s law in ic 342 may be a general case for spiral galaxies ( [ galaxies_literatures ] )
. 5 . the varied slopes from @xmath171.4 in the low-@xmath85 domain to @xmath17 2
3 in the central region may indicate the change in the main mechanism of star formation among the sub - regions in ic 342 , that is , the star formation is triggered by gravitational instability in the disk ; at the galactic center , the combination of gravitational instability and cloud - cloud collisions is favored ( [ sec_grav ] and [ sec_ccc ] ) .
the derived variable k
s law in a single galaxy also matches the theoretical prediction that the gmc properties change at the @xmath85 of approximately 85 m@xmath25 pc@xmath26 and star formation efficiency increases with @xmath85 ( [ sec_twocomp ] ) .
we would like to thank the anonymous referee for comments that helped to improve the manuscript .
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 .
99 aalto , s. , beswick , r. , jtte , e. 2010 , , 522 , 59 aniano g. , draine b. t. , gordon k. d. , sandstrom k. , 2011 , , 123 , 1218 arimoto n. , sofue y. , & tsujimoto t. , 1996 , , 48 , 275 becklin , e. e. et al .
1980 , , 236 , 441 bergin , e. a. , snell , r. l. , & goldsmith , p. f. , 1996 , , 460 , 343 bernard , j .- ph . et al .
2010 , , 518l , 88 bigiel , f. et al .
2008 , , 136 , 2846 calzetti , d. et al .
2005 , , 633 , 871 calzetti , d. et al .
2007 , , 666 , 870 calzetti , d. et al .
2010 , , 714 , 1256 calzetti d. , 2012 , preprint ( arxiv:1208.2997 ) calzetti , d. , liu , g. , & koda , j. 2012 , , 752 , 98 cedrs , cepa , j. , bongiovanni , . ,
castaeda , h.;,snchez - portal , m. , & tomita , a. 2012 , , 545 , 43 cox , p. , & mezger , p. g. 1989 , , 1 , 49 crosthwaite , l. p. , turner , j. t. , & ho p. t. p. 2000 , , 119 , 1720 crosthwaite , l. p. et al .
2001 , , 122 , 797 decarli , r. et al .
2012 , , 752 , 2 dickman , r. l. , snell , r. l. , & schloerb , f. p. 1986
, , 309 , 326 dobbs , c. l. & pringle , j. e. 2009 , , 396 , 1579 downes , d. , radford , s. j. e. , guilloteau , s. , guelin , m. , greve , a. & morris , d. 1992 , , 262 , 424 elmegreen , b. g. 1994 , , 425 , l73 elmegreen , b. g. 2002 , , 577.206 espada , d. et al .
2010 , , 720 , 666 feldmann , r. , gnedin , n. y. , & kravtsov , a. v. 2011 , apj , 732 , 115 galametz , m. et al .
2012 , , 425 , 763 gao , y. , & solomon , p. m. 2004 , , 152 , 63 genzel , r. et al .
2012 , , 746 , 69 gordon karl d. et al .
2008 , , 682 , 336 goldsmith , paul f. 2001 , , 557 , 736 goldsmith , p. f. , heyer , m. , narayanan , g. , snell , r. , li , d. , & brunt , c. 2008 , , 680 , 428 henkel , c. & mauersberger , r. , 1993 , , 274,730 henkel , c. , chin , y .-
n . ; mauersberger , r. , whiteoak , j. b. 1998 , , 329 , 443 kainulainen , j. , beuther , h. , henning , t. & plume , r. 2009 , , 508l , 35 koda , j. & sofue , y. 2006 , , 299 , 312 kravtsov , a. v. 2003 , , 590l , 1 krumholz , m. r. , mckee , c. f. , & tumlinson , j. 2008 , , 689 , 865 krumholz , m. r. , mckee , c. f. & tumlinson , j. 2009 , , 699 , 850 kuno , n. et al .
2007 , , 59 , 117 leroy , a. k. , walter , f. , brinks , e. , bigiel , f. , de blok , w. j. g. , madore , b. ; thornley , m. d. 2008 , , 136 , 2782 li , a. & draine , b. t. , 2001 , , 554 , 778 li , y. , mac low , m .- m & klessen , r. s. 2006 , , 639 , 879 liu , g. et al .
2011 , , 735 , 63 liu , l. , & gao , y. 2012 , scchg , 55 , 347 markwardt c. b. , 2009 , 2009 , aspc , 411 , 251 martel , h. , urban , a. , evans , n. j. , ii 2012 , , 757 , 59 mccall , m.l . , rybski , p.m. & shields g.a .
1985 , , 57 , 1 mcquinn , k. b. w.et al .
2002 , , 576 , 274 moustakas , j. , kennicutt , r. c. jr .
, tremonti , c. a. , et al .
2010 , ajs , 190 , 233 nakai , n. , & kuno , n. 1995 , , 47 , 761 narayanan , desika , krumholz , mark r. , ostriker , eve c. , & hernquist , lars 2012 , , 421 , 3127 narayanan , d. , krumholz , m. r. , ostriker , e. c. , & hernquist , l. 2011 , , 418 , 664 onodera , s. et al .
, 2010 , 722l , 127 paglione , t. a. d. , wall , w. f. , young , j. s. , et al .
2001 , , 135 , 183 pineda , j. l. et al .
2008 , , 679 , 481 pineda , j. l. et al .
2010 , , 721 , 686 pilyugin , l. s. 2000 , , 362 , 325 pilyugin , l. s. 2001a , , 369 , 594 pilyugin , l. s. 2001b , , 373 , 56 pilyugin , l. s. ; vilchez , j. m. ; contini , t. 2004 , , 425 , 849 pilyugin , leonid s. ; thuan , trinh x. 2005 , , 631 , 231 pirogov , l. et al .
2003 , , 405 , 639 popescu , c. c. et al .
2002 , , 567 , 221 rahman , n. et al .
2011 , , 730 , 72 rand , r. j. & kulkarni , s. r. 1990 , , 349 , 43 rieke , g. h. , et al . 2004 , , 154 , 25 rebolledo , d. et al .
2012 , , 757 , 155 saha , a. , claver , j. , hoessel , j. g.2002 , , 124 , 839s schinnerer , e. , bker , t. , & meier , d. s. 2003 , , 591l , 115 schmidt , m. 1959 . , 129,243 schruba , a. , leroy , a. k. , walter , f. , sandstrom , k. , & rosolowsky , e. 2010 , , 22 , 1699 strong , a. w. , et al .
1988 , , 207 , 1 tasker , e. j. & tan , j. c. 2009 , , 700 , 358 tan , j. c. 2000 , , 536 , 173 toomre , a. 1964 , , 139 , 1217 turner , j. l. , & ho , p. t. p. 1983
, , 268 , l79 wada , k. , & norman , c. a. 2007 , , 660 , 276 watson , w. d. , anicich , v. g. , & huntress , w. t. 1976 , , 205 , l165 werk j. k. , putman m. e. , meurer g. r. , santiago - figueroa n. , 2011 , , 735 , 71 wilson , christine d. 1995 , , 448l , 97 wilson , t. l. , rohlfs , k. , & httemeister , s. 2009 , tools of radio astronomy ( berlin : springer - verlag ) wolfire , m. g. , hollenbach , d. , & mckee , c. f. 2010 , , 716 , 1191 wu , j. et al .
2005 , , 635l , 173 yang , c .- c . , gruendl , r. a. , chu , y .- h . , mac low , m .- m . , & fukui , y. 2007 , , 671 , 374 yao , lihong ; bell , t. a. ; viti , s. ; yates , j. a. ; seaquist , e. r. 2006 , , 636 , 881 region derived from @xmath3co .
the best fit with all the data points is shown as a thick solid black line with a slope of 1.78 @xmath78 0.07 .
the grey line presents the k s law with a threshold of 100 m@xmath25 pc@xmath26 .
the slope of the grey line is 2.27 @xmath78 0.21 .
the k s relation with low @xmath5 ( @xmath0co , @xmath2 ) is overlaid with a thin solid black line .
, scaledwidth=44.0% ] l*6cr & @xmath0co ( @xmath4 ) & @xmath0co ( @xmath2 ) & @xmath3co + + low @xmath5 ( all data ) & 1.11 @xmath78 0.02 & 1.39 @xmath78 0.03 & + high @xmath5 ( all data ) & 1.33 @xmath78 0.02 & 1.94 @xmath78 0.11 & 1.78 @xmath78 0.07 + high @xmath5 ( @xmath19 100 m@xmath25 pc@xmath26 ) & 1.96 @xmath78 0.16 & 2.65 @xmath78 0.22 & 2.27 @xmath78 0.21 + + low @xmath5 ( disk ) & 0.96 @xmath78 0.06 & 1.35 @xmath78 0.07 & + high @xmath5 ( starburst ) & 1.88 @xmath78 0.14 & 2.32 @xmath78 0.13 & + | the kennicutt - schmidt ( k s ) law in ic 342 is examined using the @xmath0co - to - h@xmath1 conversion factor ( @xmath2 ) , which depends on the metallicity and co intensity .
additionally , an optically thin @xmath3co ( 1 - 0 ) is also independently used to analyze the k s law .
@xmath2 is two to three times lower than the galactic standard @xmath4 in the galactic center and approximately two times higher than @xmath4 at the disk .
the surface densities of molecular gas ( @xmath5 ) derived from @xmath0co and @xmath3co are consistent at the environment in a high-@xmath5 region . by comparing the k - s law in the disk and the central regions of ic 342
, we found that the power law index of k - s law ( @xmath6 ) increases toward the central region .
furthermore , the dependence of @xmath6 on @xmath5 is observed .
specifically , @xmath6 increases with @xmath5 .
the derived @xmath6 in this work and previous observations are consistent with the implication that star formation is likely triggered by gravitational instability in the disk ( low-@xmath5 region ) of ic 342 and both gravitational instability and cloud - cloud collisions in the central region ( high-@xmath5 regime ) .
in addition , the increasing @xmath6 toward the high-@xmath5 domain also matches the theoretical prediction regarding the properties of giant molecular clouds .
the results of ic 342 are supported by the same analysis of other nearby galaxies . |
chiral effective field theory ( eft ) as proposed in the pioneering works of weinberg @xcite is a powerful tool for the derivation of baryonic forces . in this scheme
there is an underlying power counting which allows one to improve calculations systematically by going to higher orders in a perturbative expansion . in addition , it is possible to derive two- and corresponding three - body and even four - body forces as well as external current operators in a consistent way . for a review ,
see e.g. ref .
@xcite .
recently , a hyperon - nucleon ( @xmath5 ) interaction has been derived up to next - to - leading order ( nlo ) in chiral eft by the jlich - bonn - munich group @xcite . at that order
there are contributions from one- and two - pseudoscalar - meson exchange diagrams ( involving @xmath6 , @xmath7 , @xmath8 ) and from four - baryon contact terms without and with two derivatives .
su(3 ) flavor symmetry is imposed for constructing the @xmath5 interaction .
the assumed su(3 ) symmetry allows us to fix all the coupling constants at the various meson - baryon vertices , and it reduces the number of free parameters which arise within the eft approach in form of low - energy constants ( lecs ) associated with the aforementioned contact terms . in the actual calculation the su(3 ) symmetry is broken , however , because the physical masses of the involved mesons and octet baryons ( @xmath9 , @xmath2 , @xmath3 , @xmath10 ) are used . with this interaction an excellent reproduction of available low - energy @xmath11 and @xmath12 scattering data could be achieved as reflected in a total @xmath13 of about @xmath14 for the @xmath15 data points included in the fit @xcite .
thus , the description of the @xmath5 system at nlo is on the same level of quality as by the most advanced meson - exchange @xmath5 interactions . in the present work
we investigate the properties of our @xmath5 interactions in nuclear matter .
specifically , we report results for the single particle ( s.p . )
potential of the @xmath2 and @xmath3 hyperons in nuclear matter obtained in a conventional @xmath4-matrix calculation based on the standard ( gap ) choice for the intermediate spectrum . of specific interest for us are two long - standing issues that have been discussed extensively in the literature , namely the repulsive nature of the @xmath3-nucleus potential and the weak @xmath2-nucleus spin - orbit interaction . for recent overviews of strangeness nuclear physics
see refs .
@xcite .
it is now generally accepted that the @xmath3-nuclear potential is repulsive @xcite .
phenomenological evidence comes , for example , from recently measured @xmath16 inclusive spectra related to @xmath17-formation in heavy nuclei @xcite which suggest a repulsive @xmath3-nucleus potential @xcite .
but also the theoretical interpretation of measured level shifts and widths of @xmath17 atoms @xcite requires a repulsive component of the @xmath3-nucleus potential .
microscopic models of the @xmath5 interaction , fitted to @xmath0 and @xmath1 scattering data , often fail to produce a repulsive @xmath3-nuclear potential .
specifically , for models based on meson - exchange dynamics it is rather difficult to obtain such a repulsion as is witnessed by results of @xmath4-matrix calculations for various generations of the nijmegen @xmath5 potential @xcite .
only the most recent version @xcite could overcome this difficulty @xcite - but at the expense of a noticeable increase in the achieved @xmath13 .
the situation is different for @xmath5 potentials derived within a constituent quark model @xcite . here , the interaction in the @xmath18 partial wave of the @xmath19 @xmath12 channel is always strongly repulsive due to pauli blocking effects on the quark level @xcite , and as direct consequence of this feature a repulsive @xmath3-nuclear potential is obtained in pertinent @xmath4-matrix calculations .
interestingly , our leading order ( lo ) @xmath5 interaction , published in 2006 @xcite , predicted also a repulsive @xmath18 partial wave .
calculations of the @xmath3 s.p .
potential , performed by kohno with a low - energy equivalent representation @xcite , revealed that this interaction produces indeed a repulsive @xmath3-nuclear potential .
the lo interaction involves only five contact terms and the lecs associated with them were determined in a fit to the low - energy @xmath0 and @xmath1 cross sections .
no attention was paid to the sign of the @xmath19 @xmath12 amplitude in the @xmath18 partial wave .
in the course of constructing the eft interaction @xcite up to nlo ( which involves altogether 13 lecs in the @xmath20 waves and the @xmath20@xmath21 transitions ) it turned out that the available @xmath5 scattering data could be fitted equally well with an attractive or a repulsive interaction in the @xmath18 partial wave of the @xmath19 @xmath12 channel .
for the reasons discussed above , the repulsive solution was adopted .
the resulting interaction produces a moderately repulsive @xmath18 phase shift as can be seen in fig .
9 of ref .
first @xmath4-matrix calculations for this interaction will be reported in the present work .
it turns out that a repulsive @xmath3-nuclear potential is predicted .
hypernuclear spectra provide clear indication for a rather weak @xmath2-nucleus spin - orbit potential @xcite . apparently , also this feature is difficult to reproduce with microscopic models of the @xmath5 interaction .
in particular , meson - exchange potentials tend to grossly overestimate the spin - orbit force , judging from applications of various nijmegen models in the studies presented in refs .
@xcite and @xcite .
calculations based on @xmath5 interactions derived in the quark model lead to ( partially ) more promising results @xcite . according to the discussion of the strength of the spin - orbit potential in refs .
@xcite , done on the level of the scheerbaum factor @xcite calculated from the @xmath4-matrix , there is a sizeable antisymmetric spin - orbit component generated from the quark model which cancels to a large extent the symmetric component in the @xmath0 isospin @xmath22 channel so that an overall weak spin - orbit potential is achieved .
the aforementioned work by kohno @xcite provides also results for the scheerbaum factor for our lo @xmath5 potential .
the predicted value was found to be fairly small and somewhat surprisingly of opposite sign as compared to results from standard potential models . at leading order
there is no contact term that would give rise to an antisymmetric spin - orbit component .
thus , the result at lo is a pure prediction based on the dynamics generated by pseudoscalar meson exchange .
this is different at nlo where such a contact term occurs and facilitates @xmath23@xmath24 transitions in the coupled ( @xmath22 ) @xmath0@xmath1 system @xcite . since the corresponding lec could not be pinned down by a fit to the existing @xmath0 and @xmath1 scattering data it was simply put to zero in ref .
@xcite . however , it is interesting to see whether that lec can be fixed now from a study of the properties of the @xmath25 hyperon in nuclear matter and , specifically , whether it can be utilized to achieve a weak @xmath2-nuclear spin - orbit potential .
this is indeed the case as will be demonstrated in the present work .
note that the same issues have been addressed by kaiser and weise @xcite within an approach similar to ours , but on a perturbative level .
it was found that the iterated ( second order ) one - pion exchange ( with @xmath25 and @xmath26 intermediate states ) gives rise to a repulsive @xmath26-nuclear potential .
note that dimensional regularization has been employed to extract the genuine long - range components .
second - order pion exchange plays also a decisive role for the spin - orbit force , where it is argued that its contribution counterbalances the short - range components due to the scalar and vector mean fields to a large extent in case of the @xmath25 and , accordingly , leads to a small net result for the @xmath25-nuclear spin - orbit force .
the same mechanism is at work for the @xmath3-nuclear spin - orbit coupling @xcite .
an analogous calculation where , in addition , contributions involving the decuplet baryons @xmath27 and @xmath28 were considered @xcite led to the same conclusions concerning the @xmath25-nuclear spin - orbit force .
the paper is structured as follows : in sect .
2 the formalism employed for evaluating the @xmath4-matrix is outlined .
3 provides a basic introduction to our eft @xmath5 potentials .
specifically , the implementation of the antisymmetric spin - orbit force and its impact on the @xmath0 and @xmath1 results are discussed .
results of our calculation are presented and analysed in sect .
besides those based on our chiral eft interactions , predictions for three meson - exchange potentials ( developed by the jlich and nijmegen groups , respectively ) are included for illustration purposes .
the paper ends with a short summary .
the standard formalism for the treatment of a hyperon in nuclear matter is the conventional brueckner theory . for convenience
we summarize below the essential formulae employed in our calculation .
a more detailed description can be found in ref .
@xcite , see also @xcite .
we consider a @xmath2 or @xmath3 hyperon with momentum @xmath29 in nuclear matter of density @xmath30 , where @xmath31 denotes the fermi momentum of nuclear matter . in order to determine the properties of those hyperons we employ the brueckner reaction - matrix formalism as discussed in refs .
the @xmath5 reaction matrix , @xmath32 , is defined by the bethe - goldstone equation @xmath33 with @xmath34 , @xmath35 = @xmath25 , @xmath26 . here , @xmath36 is the pauli projection operator , which excludes those intermediate @xmath5 states with the nucleon inside the fermi sea .
the starting energy @xmath37 for an initial @xmath5 state with momenta @xmath29 , @xmath38 is given by @xmath39 where the s.p .
( @xmath41 ) includes not only the ( nonrelativistic ) kinetic energy and the particle mass but in addition the s.p .
potential @xmath42 : @xmath43 here and in the following , @xmath44 denotes the modulus of @xmath29 , etc . in practice ,
the bethe - goldstone equation is solved for a given @xmath34 momentum @xmath29 and relative momentum @xmath45 of the initial @xmath11 or @xmath12 state . for that purpose
we re - write the energy denominator in eq .
( [ eq : g1 ] ) in the form @xmath46 where we have introduced the total and relative momenta , @xmath47 and @xmath48 , which are related to the @xmath34 and @xmath9 momenta @xmath49 and @xmath50 via @xmath51 the reduced mass in eq .
( [ eq:3 ] ) is given by @xmath52 .
the square of the ( conserved ) total momentum , @xmath53 , can be written as @xmath54 with @xmath55 .
note that @xmath56 depends via @xmath57 on the angle between @xmath29 and @xmath45 .
also the pauli projection operator , @xmath58 depends on the angle between the total momentum @xmath47 and the relative momentum @xmath48 of the intermediate @xmath5 state . in order to facilitate a partial - wave decomposition ( with regard to the total @xmath5 angular momentum @xmath59 , with @xmath59 the sum of the total spin @xmath60 and the relative angular momentum @xmath61 : @xmath62 ) it is common practice to approximate @xmath57 and @xmath63 by their angle - averaged values , evaluated for given @xmath44 , @xmath64 of the initial @xmath5 state .
these are given by @xmath65 ^ 2 ] , \ \
\rm{otherwise}~ , \\
\end{array } \right .
\end{aligned}\ ] ] @xmath66 / 4k\sqrt{\overline{k^2 } } , \ \
\rm{otherwise } ~.\\ \end{array } \right.\ ] ] when evaluating the starting energy @xmath67 the nucleon momentum of the initial @xmath5 state , @xmath68 , is approximated by @xmath69^{1/2 } \ .\ ] ] the partial - wave expanded bethe - goldstone equation reads @xmath70 where @xmath71 stands for @xmath2 or @xmath3 and @xmath72 is the total @xmath5 isospin which can be @xmath73 or @xmath74 .
the hyperon s.p .
potential @xmath75 is calculated from @xmath76 where @xmath77 .
the weight function @xmath78 is given by @xmath79 ^ 2 \over 4\xi_y ( 1+\xi_y)p_y p } , \ \
\rm{otherwise}~. \\
\end{array } \right.\ ] ] we evaluate the @xmath2 and @xmath3 s.p .
potentials @xmath80 and @xmath81 self - consistently in the standard way for a fixed fermi momentum @xmath31 . in case of the @xmath26 hyperon
we decompose the s.p .
potential into an isoscalar and an isovector based on the parameterization @xcite @xmath82 here , @xmath83 and @xmath84 are the isospin operators of the @xmath3 and the nucleus ( with @xmath85 nucleons ) .
the isoscalar and isovector components are given by @xmath86 and @xmath87 , respectively , where @xmath88 are the corresponding parts of the sum over @xmath72 in eq .
( [ jh02 ] ) . in case of the @xmath26 hyperon , the propagator in eq .
( [ eq:3 ] ) can become singular if the @xmath26 s.p .
potential is repulsive .
we avoided the proper but technically involved treatment of this singularity by adding a small finite imaginary part in our calculation .
we verified that this prescription has no influence on the obtained results for the real part of the @xmath26 s.p .
however , we can not calculate the imaginary part of @xmath81 . in any case
, it is known that the imaginary part can be only determined reliably when using the continuous choice for the intermediate spectrum @xcite and we plan to perform such a calculation in the future @xcite . in the present work ( like before in @xcite ) , the nucleon s.p .
potential @xmath89 is taken from a calculation @xcite of pure nuclear matter employing a phenomenological @xmath90 potential .
it has been found that the results for the @xmath2 and @xmath3 hyperon in nuclear matter for the gap choice are not too sensitive to the choice of @xmath89 @xcite .
nonetheless , also this limitation will be avoided in the future calculation @xcite . in order to relate the strength of the @xmath2-nuclear spin - orbit potential to the two - body @xmath11 interaction we consider the scheerbaum factor @xmath91 @xcite calculated in nuclear matter . in the scheerbaum approximation
, the @xmath2- or @xmath3-nuclear spin - orbit potential is represented by @xmath92 with @xmath93 the nucleon density distribution and @xmath94 the single - particle orbital angular momentum operator .
we adopt here the definition as given in ref .
the explicit expression for the scheerbaum factor @xmath91 derived in this reference in terms of the @xmath4-matrix elements given in momentum space reads in our notation @xmath95 \right\}\ \ .\nonumber \\
\label{sb2}\end{aligned}\ ] ]
a comprehensive description of the derivation of the chiral baryon - baryon potentials for the strangeness sector using the weinberg power counting can be found in refs .
the lo potential consists of four - baryon contact terms without derivatives and of one - pseudoscalar - meson exchanges while at nlo contact terms with two derivatives arise , together with contributions from ( irreducible ) two - pseudoscalar - meson exchanges .
the contributions from pseudoscalar - meson exchanges ( @xmath6 , @xmath8 , @xmath7 ) are completely fixed by the assumed su(3 ) flavor symmetry in terms of the @xmath96 and @xmath21 couplings . on the other hand ,
the strength parameters associated with the contact terms , the lecs , need to be determined by a fit to data .
how this is done is described in detail in ref .
@xcite and , therefore , we will be brief here .
since su(3 ) symmetry is also imposed for the contact terms , the number of independent lecs that can contribute is significantly reduced . in particular , for the @xmath0@xmath1 system there are in total 13 lecs entering the @xmath20-waves and the @xmath20@xmath21 transitions , respectively , and 10 constants in the @xmath97-waves .
the @xmath20-wave lecs could be fixed by a fit to available low - energy total cross sections for the reactions @xmath98 , @xmath99 , @xmath100 , @xmath101 , and @xmath102 .
information from the nucleon - nucleon ( @xmath90 ) sector was not needed .
the limited number ( and quality ) of differential cross sections and the complete lack of polarization observables makes a determination of the contact terms in the @xmath97-waves from @xmath0@xmath1 data alone impossible .
therefore , in this case @xmath90 @xmath97-wave phase shifts had to be used as a further constraint for the contact terms .
the lec that generates the antisymmetric spin - orbit force , and which produces a @xmath23-@xmath24 transition , was set to zero because it could not be determined from the @xmath5 scattering data .
the reaction amplitudes are obtained from the solution of a coupled - channel lippmann - schwinger ( ls ) equation for the interaction potentials @xcite .
the potentials in the ls equation are cut off with a regulator function , @xmath103 $ ] , in order to remove high - energy components @xcite . in ref .
@xcite cutoff values in the range @xmath104 @xmath105mev are considered , similar to what was used for chiral @xmath90 potentials @xcite .
the variation of the results with the cutoff can be viewed as a rough estimate for the theoretical uncertainty @xcite . as demonstrated in ref .
@xcite , the available @xmath11 and @xmath12 scattering data are very well described by the eft interaction .
indeed the description of the @xmath5 system achieved at nlo is on the same level of quality as the one by the most advanced meson - exchange @xmath5 interactions . besides an excellent description of the @xmath5 data the chiral eft interaction yields a satisfactory value for the hypertriton binding energy ,
see ref .
calculations for the four - body hypernuclei @xmath106 and @xmath107 based on the eft interactions can be found in ref .
@xcite .
let us come back to the antisymmetric spin - orbit force .
the general structure of the spin- and momentum - dependence of the contact terms at next - to - leading order is given by @xcite @xmath108 where @xmath109 ( @xmath110 ) are the lecs .
the transferred and average momenta , @xmath111 and @xmath48 , are defined in terms of the final and initial center - of - mass momenta of the baryons , @xmath112 and @xmath45 , as @xmath113 and @xmath114 . the term proportional to @xmath115 in eq .
( [ vc ] ) represents an antisymmetric spin - orbit force and gives rise to ( spin ) singlet - triplet ( @xmath116 ) transitions . after performing a partial - wave projection and imposing su(3 ) symmetry constraints the @xmath23-@xmath24 transition potentials read in the notation of ref .
@xcite @xmath117 with @xmath118 and @xmath119 .
expressions for the time - reversed transitions follow from the requirement of time - reversal invariance .
the superscript @xmath120 of the lec is a reminder that the singlet - triplet transition is accompanied by a transition between the symmetric and antisymmetric su(3 ) octet representations as required by the generalized pauli principle .
note that , in line with ref .
@xcite , we have absorbed here the overall factor @xmath121 ( see table 1 in ref .
@xcite ) into the definition of @xmath122 . in the present work we fix the lec @xmath122 by considering the scheerbaum factor @xmath123 @xcite calculated from the @xmath4-matrix . with regard to its `` empirical '' value we took studies of the splitting of the @xmath124 and @xmath74 states of @xmath125 by hiyama et al . @xcite and fujiwara et al .
@xcite as guideline .
the results of those authors suggest that values for @xmath123 in the order of @xmath126 to @xmath127 mev @xmath128 would be needed to reproduce the experimentally observed small level splitting @xcite .
values of around @xmath129 and @xmath130 mev @xmath128 are suggested in refs .
@xcite and @xcite , respectively . since the precise value required for the @xmath2 s.p .
spin - orbit strength can only be pinned down via a dedicated calculation of finite hypernuclei based on our eft interactions , which is beyond the scope of the present study , we decided to aim at a typical result of @xmath131 mev @xmath128 .
indeed , any reference value at the expected order of magnitude would be sufficient for the main goal of the present investigation , namely for exploring in how far the lec @xmath122 can be used to achieve a small @xmath2 s.p .
spin - orbit potential .
@xmath23 and @xmath24 phase shifts and the @xmath23@xmath24 mixing parameter @xmath132 as a function of @xmath133 .
the red / dark band shows the chiral eft results to nlo for variations of the cutoff in the range @xmath134 500,@xmath135,650 mev , while the green / light band are results to lo for @xmath134 550,@xmath135,700 mev .
the hatched band is the nlo fit from @xcite without antisymmetric spin - orbit force .
the dashed curve is the result of the jlich 04 meson - exchange potential @xcite . , title="fig:",height=241 ] @xmath23 and @xmath24 phase shifts and the @xmath23@xmath24 mixing parameter @xmath132 as a function of @xmath133 .
the red / dark band shows the chiral eft results to nlo for variations of the cutoff in the range @xmath134 500,@xmath135,650 mev , while the green / light band are results to lo for @xmath134 550,@xmath135,700 mev .
the hatched band is the nlo fit from @xcite without antisymmetric spin - orbit force .
the dashed curve is the result of the jlich 04 meson - exchange potential @xcite . , title="fig:",height=241 ] @xmath23 and @xmath24 phase shifts and the @xmath23@xmath24 mixing parameter @xmath132 as a function of @xmath133 .
the red / dark band shows the chiral eft results to nlo for variations of the cutoff in the range @xmath134 500,@xmath135,650 mev , while the green / light band are results to lo for @xmath134 550,@xmath135,700 mev .
the hatched band is the nlo fit from @xcite without antisymmetric spin - orbit force .
the dashed curve is the result of the jlich 04 meson - exchange potential @xcite .
, title="fig:",height=241 ] [ tab : f1 ] .the @xmath5 contact terms for the @xmath23 partial wave and the @xmath23@xmath24 transition potential for various cut offs .
the values of the lecs are in @xmath136 @xmath137 , the values of @xmath2 in mev . [ cols="^,^,>,>,>,>,>,>",options="header " , ] [ tab : si ]
we have investigated the in - medium properties of a hyperon - nucleon potential , derived within chiral effective field theory and fitted to low - energy @xmath0 and @xmath1 scattering data .
in particular , we have evaluated the single - particle potential for the @xmath2 and @xmath3 hyperons in nuclear matter in a conventional @xmath4-matrix calculation , and the scheerbaum factor associated with the hyperon - nucleus spin - orbit interaction .
results are presented for a leading - order interaction published in 2006 which accounts well for the bulk properties of the @xmath0 and @xmath1 system @xcite and for our recent @xmath5 interaction derived up to next - to - leading order in chiral eft which provides an excellent description of the available @xmath5 data @xcite .
the predictions for the @xmath2 single - particle potential turned out to be in good qualitative agreement with the empirical values inferred from hypernuclear data .
a depth of about @xmath138 mev is found for the nlo interaction and of about @xmath139 mev for the lo potential .
the @xmath3-nuclear potential is found to be repulsive , in agreement with phenomenological information , with values around @xmath140@xmath141 mev .
empirical information suggests that the @xmath2-nucleus spin - orbit interaction should be rather weak .
therefore , we discussed also the spin - orbit interaction and , in particular , the role of the antisymmetric spin - orbit force in the @xmath5 system .
the chiral eft approach yields a potential that contains , besides pseudoscalar meson exchanges ( @xmath6 , @xmath7 , @xmath8 ) , a series of contact interactions with an increasing number of derivatives . in this approach
a contact term representing an antisymmetric spin - orbit force arises already at nlo .
it induces @xmath23@xmath24 transitions in the coupled ( @xmath22 ) @xmath0@xmath1 system .
the low - energy constant associated with the contact term could not be pinned down by a fit to the existing @xmath0 and @xmath1 scattering data as found in @xcite and , therefore , it was simply put to zero in that work . as demonstrated in the present study ,
it can be fixed , however , from investigating the properties of the @xmath25 hyperon in nuclear matter and , specifically , it can be utilized to achieve a weak @xmath2-nuclear spin - orbit potential . in any case , and especially with regard to the spin - orbit force , one has to acknowledge that nuclear matter calculations can provide only a first glimpse on the properties of the @xmath5 interaction in the medium .
a stringent test of the properties of the eft interactions in the medium can be only performed by deducing effective interactions ( from the @xmath4-matrix in nuclear matter ) and applying them in calculations of finite hypernuclei .
the authors acknowledge helpful communications with avraham gal , norbert kaiser , michio kohno and yasuo yamamoto . this work is supported in part by the dfg and the nsfc through funds provided to the sino - german crc 110 `` symmetries and the emergence of structure in qcd '' and by the eu integrated infrastructure initiative hadronphysics3 .
part of the numerical calculations has been performed on the supercomputer cluster of the jsc , jlich , germany .
m. kohno , y. fujiwara , y. watanabe , k. ogata and m. kawai , prog .
* 112 * ( 2004 ) 895 .
m. kohno , y. fujiwara , y. watanabe , k. ogata and m. kawai , phys .
c * 74 * ( 2006 ) 064613 .
batty , e. friedman and a. gal , phys .
b * 335 * ( 1994 ) 273 .
j. mare , e. friedman , a. gal and b.k .
jennings , nucl .
phys . a * 594 * ( 1995 ) 311 .
t. a. rijken , v. g. j. stoks and y. yamamoto , phys .
c * 59 * ( 1999 ) 21 .
y. fujiwara , y. suzuki and c. nakamoto , prog .
* 58 * ( 2007 ) 439 .
m. kohno , y. fujiwara , t. fujita , c. nakamoto and y. suzuki , nucl .
phys . a * 674 * ( 2000 ) 229 .
h. polinder , j. haidenbauer and u .- g .
meiner , nucl .
a * 779 * ( 2006 ) 244 .
m. kohno , phys .
c * 81 * ( 2010 ) 014003 .
e. hiyama , m. kamimura , t. motoba , t. yamada and y. yamamoto , phys . rev .
* 85 * ( 2000 ) 270 .
d. j. millener , nucl .
a * 835 * ( 2010 ) 11 .
d. j. millener , j. phys .
* 312 * ( 2011 ) 022005 .
y. fujiwara , m. kohno and y. suzuki , mod .
a * 24 * ( 2009 ) 1031 .
j. hu , e. hiyama and h. toki , phys . rev .
c * 90 * ( 2014 ) 014309 .
d. j. millener , c. b. dover and a. gal , phys .
c * 38 * ( 1988 ) 2700 .
y. yamamoto , h. bando and j. ofka , prog .
phys . * 80 * ( 1988 ) 757 . | the in - medium properties of a hyperon - nucleon potential , derived within chiral effective field theory and fitted to @xmath0 and @xmath1 scattering data , are investigated .
results for the single - particle potentials of the @xmath2 and @xmath3 hyperons in nuclear matter are reported , based on a conventional @xmath4-matrix calculation .
the @xmath3-nuclear potential is found to be repulsive , in agreement with phenomenological information .
a weak @xmath2-nuclear spin - orbit interaction can be achieved by an appropriate adjustment of a low - energy constant corresponding to an antisymmetric @xmath0@xmath1 spin - orbit interaction that arises at next - to - leading order in the effective field theory approach .
@xmath5 interaction , chiral effective field theory , @xmath4-matrix , hyperon single - particle potential , spin - orbit interaction |
99 p. huovinen , nucl - th/0305064 ; p. f. kolb and u. heinz , nucl - th/0305084 .
kolb , p. huovinen , u. heinz , and h. heiselberg , phys .
b * 500 * , 232 ( 2001 ) ; p. huovinen , p.f
. kolb , u.w .
heinz , p.v .
ruuskanen , and s.a .
voloshin , _ ibid .
_ b * 503 * , 58 ( 2001 ) . t. hirano , phys . rev .
c * 65 * , 011901 ( 2002 ) ; t. hirano and k. tsuda , phys .
c * 66 * , 054905 ( 2002 ) .
m. gyulassy , i. vitev , x. n. wang , and b. w. zhang , nucl - th/0302077 ; a. kovner and u. a. wiedemann , hep - ph/0304151 . s. s. adler _ et al .
_ , phenix collaboration , phys .
* 91 * , 072301 ( 2003 ) ; k. adcox _ et al .
_ , star collaboration , phys .
b * 561 * , 82 ( 2003 ) . c. adler _ et al .
_ , star collaboration , phys .
* 90 * , 082302 ( 2003 ) .
b. b. back _ et al .
_ , phobos collaboration , phys .
lett . * 91 * , 072302 ( 2003 ) ; s. s. adler _ et al .
_ , phenix collaboration , _ ibid .
_ , * 91 * , 072303 ( 2003 ) ; j. adams _ et al .
_ , star collaboration , _ ibid .
_ , * 91 * , 072304 ( 2003 ) ; i. arsene _ et al .
_ , brahms collaboration , _ ibid .
_ , * 91 * , 072305 ( 2003 ) .
r. baier , a. kovner , and u. a. wiedemann , phys .
d * 68 * , 054009 ( 2003 ) ; j. l. albacete , n. armesto , a. kovner , c. a. salgado and u. a. wiedemann , hep - ph/0307179 ; d. kharzeev , y. v. kovchegov and k. tuchin , phys
. rev .
d * 68 * , 094013 ( 2003 ) ; j. jalilian - marian , y. nara and r. venugopalan , phys .
b * 577 * , 54 ( 2003 ) .
j. p. blaizot , f. gelis and r. venugopalan , hep - ph/0402256 ; hep - ph/0402257 ; e. iancu , k. itakura , d. n. triantafyllopoulos , hep - ph/0403103 . | hadron spectra in au+au collisions at rhic are calculated by hydrodynamics with initial conditions from the color glass condensate ( cgc ) .
minijet components with parton energy loss in medium are also taken into account by using parton density obtained from hydrodynamical simulations .
we found that cgc provides a good initial condition for hydrodynamics in au+au collisions at rhic .
high energy heavy ion collisions involve different aspects according to the relevant energy or time scale .
there already exist many theoretical approaches to understand numerous rhic data . in this work
, we consider in particular physics of gluon saturation , hydrodynamic evolution , and the energy loss of hard partons in the medium .
our goals are to combine them and to take a step to a unified understanding of the dynamical aspect of high energy heavy ion collisions . at soft region
@xmath0 gev / c , where bulk dynamics is governed , hydrodynamical description @xcite is successful in describing the elliptic flow at low @xmath1 , up to semi - central collisions , and mid - rapidity @xcite at rhic .
this is one of the strongest indications of an early thermalization of the quark gluon plasma ( qgp ) at rhic .
hydrodynamics also predicts that the scaled elliptic flow , which is defined as the second harmonics @xmath2 divided by initial spatial eccentricity @xmath3 , becomes almost constant around 0.2 @xcite .
the experimental data reaches the hydrodynamic limit for the first time in central and semi - central collisions at rhic energies @xcite . on the other hand ,
minijets go across the expanding matter and lose their energies ( jet quenching ) in heavy ion collisions @xcite . observed large suppression of hadron spectra @xcite and disappearance of the away - side peak in azimuthal correlation functions at mid - rapidity @xcite
have been interpreted as a consequence of parton energy loss in the medium .
indeed , cronin enhancement of the hadron spectra and existence of the back - to - back correlation at mid - rapidity in @xmath4 collisions at rhic @xcite support the importance of the strong final state interaction in @xmath5 collisions .
the current rhic data strongly suggest that the initial parton density is large .
what is an origin of the large density in au+au collisions at rhic ?
the bulk particle production is dominated by the small @xmath6 modes in the nuclear wave function , where @xmath6 is a momentum fraction of the incident particles .
it is well known that gluon density increases rapidly with decreasing @xmath6 by the bfkl cascade until gluons begin to overlap in phase space where nonlinear interaction becomes important @xcite .
these gluons form the colour glass condensate ( cgc ) @xcite .
remarkably , the cgc results on the global observables such as the centrality , rapidity and the energy dependences of charged hadron multiplicities agree with the rhic data @xcite .
it has been shown that the classical wave function in the mv model contains cronin enhancement and the quantum evolution in @xmath6 makes the spectrum suppressed @xcite . from the above considerations ,
the cgc , hydrodynamics , and the energy loss of hard partons are key ingredients to describe the rhic physics and must be closely related with each other .
for example , the cgc could be a good initial condition for thermalization because it produces a large number of gluons .
thus these gluons are responsible to the large suppression of jet spectra . in this work
we assume that the origin of thermalized partonic matter is the cgc in high energy heavy ion collisions , and use it as an initial condition in the hydro+jet model @xcite . with this approach
we expect to get deeper understanding of the dynamical aspect of the heavy ion collisions .
in addition , some of the problems which are inherent in a particular approach can be removed .
we employ the @xmath7 factorized formula along the line of kharzeev , levin , and nardi ( kln ) @xcite for the computation of the gluon rapidity distribution which is given by @xmath8 where @xmath9 with @xmath10 and @xmath1 are a rapidity and a transverse momentum of a produced gluon .
we assume that the system of initially produced gluons reaches local thermailzed state at a short time scale .
although the produced gluons will reach the thermalized state through the dissipative processes in the realistic situations , the description of non - equilibrium phenomena is beyond the scope of the present paper .
we also assume that the shape of the rapidity density distribution is not changed during the system is thermalized .
therefore , we take initial conditions from gluon distribution obtained from eq .
( [ eq : ktfac ] ) .
assuming bjorken s ansatz @xmath11 where @xmath12 is the space - time rapidity , we obtain the number or the energy density for gluons at each space - time point . the initial transverse energy per particle yields @xmath13 gev at @xmath14 from eq . ( [ eq : ktfac ] ) .
this is within a range estimated in a numerical simulation of the classical yang - mills equation @xcite .
it should be noted that the assumption of the thermalization in cgc gluon distribution is to reduce the transverse energy per particle from @xmath15 to @xmath16 within our present parameters .
the effect of the hydrodynamic afterburner is to reduce the transverse energy per particles due to @xmath17 work and yields @xmath18 gev .
the rapidity distribution becomes slightly wider .
our result supports that kln calculation @xcite which is based on the assumption of parton - hadron duality is a good approximation on the rapidity distributions . and
200 gev is compared to the phobos data @xcite .
impact parameters which correspond to @xmath19 from phobos are used in the calculations . , title="fig:",width=288 ] and 200 gev is compared to the phobos data @xcite .
impact parameters which correspond to @xmath19 from phobos are used in the calculations .
, title="fig:",width=288 ] in fig .
[ fig : dndeta200 ] , pseudorapidity distributions of charged hadrons in au + au collisions at both @xmath20 and 200 gev are compared with the phobos data @xcite .
@xmath7 factorization approach in the cgc provides very good initial conditions for the hydrodynamical simulations which reproduce rapidity , centrality and energy dependences of multiplicity .
it should be emphasized that it is very hard to find such a good initial condition which fits the data with the same quality as the cgc one presented here .
( right ) in au + au collisions at @xmath21 are compared to the phenix data @xcite . , title="fig:",width=288,height=220 ] ( right ) in au + au collisions at @xmath21 are compared to the phenix data @xcite .
, title="fig:",width=288 ] we now turn to the discussion of the high @xmath1 hadron spectrum . in our model ,
high @xmath1 jets suffer interaction with the local parton density which is governed by hydrodynamic evolution .
we only take into account parton energy loss in deconfined matter @xmath22 . in fig .
[ fig : raa45 ] , the result of the centrality dependence of the pion spectrum and the nuclear modification factor @xmath23 integrated over @xmath24 gev / c for @xmath25 are compared to phenix data @xcite .
our results only account for the data up to mid - central events and fail to reproduce data at peripheral collisions , because neither cgc nor hydrodynamics can be applied in the low density region .
the centrality dependence is well described by assuming the number of participants scaling @xcite .
however , note that our density scales as @xmath26 at mid - rapidity . by comparison
, we also plot the lopqcd calculations with the same initial condition assuming ( 1 + 1)d expansion of the system ( @xmath27 ) .
lopqcd calculation without parton energy loss at @xmath28 mev is consistent with our hydrodynamic result .
however , if there is no restriction on the minimum temperature in the calculation of energy loss , we see that agreement with data becomes somewhat better .
this may indicate the contribution from hadronic interactions at peripheral collisions . in summary , cgc , hydrodynamics and
the energy loss of hard jets have been integrated into one model and this dynamical approach describes the bulk properties of rhic data .
a systematic study within this dynamical approach is in progress .
we are grateful to j. jalilian - marian , u. heinz , d. kharzeev , l. mclerran , and r. venugopalan for useful comments .
the work of t.h . is supported by riken .
research of y.n .
is supported by the doe under contract no .
de - fg03 - 93er40792 . |
exchange of scalar or vector bosons , @xmath2 , with mass @xmath3 between two non - relativisitic fermions generically produces a potential @xmath4 where the @xmath5 and @xmath6 signs refer to scalar and vector interactions respectively , and @xmath7 . for arbitrary vector interactions between electrically neutral atoms with proton and neutron numbers @xmath8 and @xmath9 , we have @xmath10 where @xmath11 @xmath12 is an angle that , in principle , could have any value between @xmath13 and @xmath14 , and the @xmath15 s are vector `` charges '' . eqs .
[ eq : gv ] and [ eq : psi ] can also be applied to scalar interactions , even though they are not exact because scalar charges are not conserved and binding energy can carry a charge . expressing eq .
[ eq : vector ] in terms of eq . 1 , we have @xmath16_a \left[\frac{\tilde{q}}{\mu}\right]_b \right)^{-1}~ , \label{eq : gsquared}\ ] ] where @xmath17 with @xmath18 and @xmath19 being the atomic mass and atomic mass unit respectively , and @xmath20= \left[\frac{z}{\mu}\right ] \cos \tilde{\psi } + \left[\frac{n}{\mu}\right ] \sin{\tilde{\psi}}~. \label{eq : charge}\ ] ] the molybdenum pendulum and attractor used in ref .
@xcite have @xmath21\!=0.4378134 $ ] and @xmath22\!=0.5631686 $ ] . figure [ fig : alphalambda ] illustrates the upper limits implied by the results of ref .
@xcite on vector interactions coupled to @xmath23 where @xmath24 and @xmath25 are baryon and lepton numbers , respectively .
figure [ fig : theta5 ] shows how the upper limits on @xmath26 depend on the parameter @xmath27 that specifies the charge @xmath28 .
in string theories , the geometry of spacetime is expected to be dynamical with the radii of new dimensions fluctuating independently at each point in our 4-dimensional spacetime . in an effective low - energy theory ,
the volume of the extra dimensions must be stabilized by _ radions _ , low - mass spin-0 fields with gravitational - strength couplings that determine the radius of the @xmath29 `` large '' extra dimensions .
radion exchange will produce a yukawa force with a strength and range@xcite @xmath30 @xmath31 ^ 2\;{\rm mm}~ , \label{eq : radion range}\ ] ] where @xmath32 is the unification mass .
in many cases the radion - mediated force is the longest - range effect of new dimensions@xcite because it does not diminish as the number of new dimensions increases . for @xmath33 and @xmath34 ( @xmath35 and @xmath36 ) , the data of ref . @xcite
give @xmath37 string theories predict a scalar partner of the graviton , the dilaton , whose mass is initially zero .
equivalence principle experiments have ruled out a massless dilaton .
kaplan and wise@xcite evaluated the coupling of a low - mass dilaton to strongly interacting matter and shown that its coupling to matter should satisfy @xmath38 .
figure 6 of ref .
@xcite then sets a 95% confidence lower bound on the dilaton mass of @xmath39
recently , the pvlas collaboration@xcite studied the propagation of optical photons through a vacuum containing a strong transverse @xmath24 field .
they reported an optical rotation at least @xmath40 times larger than the qed prediction , and speculated that this was evidence for a new spin - zero particle that , through a second - order process , mixes with the photon in a magnetic field as shown in fig .
[ fig : pvlas diag ] . the apparent sign of the observed rotation requires the new particle to be a scalar ( as opposed to pseudoscalar ) boson@xcite , and the magnitude requires @xmath41 this @xmath42 vertex generates , by the second - order process shown in fig .
[ fig : pvlas diag ] , an effective scalar interaction between two protons , which to leading order is estimated to be@xcite @xmath43 assuming that @xmath44 is large compared to @xmath45 and @xmath46 ( _ i.e. _ @xmath47 ) , our scalar constraints ( fig .
[ fig : theta5 ] with @xmath48 ) place the upper limit shown in fig .
[ fig : alphalambda ] ; in particular @xmath49 which is inconsistent with eq .
[ eq : pvlas claim ] by a factor of @xmath50 . for @xmath51 ,
the bound in eq .
[ eq : our 2 gamma bound ] and fig .
[ fig : alphalambda ] improves on the astrophysical constraint@xcite by a factor up to @xmath52 .
( both of these bounds would be relaxed in models where the @xmath42 interaction has an additional low - energy form factor . )
chameleons are scalar fields that couple to themselves and to matter with gravitational strength@xcite .
chameleon exchange leads to an effective potential density@xcite @xmath53 where @xmath54 characterizes the strength of the self interaction , @xmath55 characterizes the coupling of the scalar field to matter , and @xmath56 is the reduced planck mass .
the `` natural '' values of @xmath55 and @xmath54 are @xmath57 . in the presence of matter with density @xmath58
, a massless chameleon field acquires an effective mass@xcite , @xmath59 that dramatically weakens experimental constraints because only a small amount of material near the surface with thickness @xmath60 contributes to a long - range force@xcite . for @xmath61 g/@xmath62 and @xmath63
, this skin thickness is about 60 @xmath64 m . using the method outlined in ref .
@xcite , we calculated the 21@xmath65 chameleon torque , @xmath66 , as a function of pendulum / attractor separation @xmath67 for the apparatus of ref .
the combined ref .
@xcite data were fitted with the newtonian torque plus @xmath66 to generate the 95% confidence level constraints on @xmath54 as a function of @xmath55 shown in fig .
[ fig : betagamma ] . a substantial region of parameter space around the natural values @xmath68 and @xmath69 is strongly excluded .
sundrum@xcite suggested a solution to the cosmological constant problem , namely that the observed `` dark energy '' density is much smaller than the vacuum energy density predicted by the usual rules of quantum mechanics .
he conjectured that this could be explained if the graviton is a `` fat '' object whose size @xmath70 prevents it from `` seeing '' the short - distance physics that dominates the vacuum energy . in his conjecture ,
the gravitational force vanishes at sufficiently small separations compared to @xmath70 .
we test this scenario by assuming that the gravitational force is @xmath71~ , \label{eq : fat graviton}\ ] ] which has a shape similar to the force shown pictorially in fig . 8 of ref . @xcite ; it vanishes at @xmath72 and has a maximum at @xmath73 .
sundrum argues that naturalness requires @xmath74 m .
our results require @xmath75 m at 95% confidence .
our upper limit on @xmath70 is larger than our @xmath76 m limit on the size of an extra dimension because our data probe the large - distance tail of the new potentials , and the `` fat - graviton '' force falls off much more rapidly with increasing separation than does a yukawa force .
we constrained power - law interactions of the form @xmath77 by fitting the combined data of ref .
@xcite with a function that contained the newtonian term and a single power - law term with @xmath78 , 3 , 4 , or 5 .
the results are listed in table [ tab : power - law constraints ] together with constraints from previous isl tests .
power - law interactions arise from higher - order exchange processes with simultaneous exchange of multiple massless bosons .
second - order processes are particularly interesting when the exchanged particles are unnatural - parity bosons ( for which the 1st - order force vanishes when averaged over unpolarized test bodies ) or fermions ( for which the 1st - order process is forbidden )
. potentials with @xmath78 are generated by the simultaneous exchange of two massless scalar@xcite bosons .
simultaneous exchange of massless pseudoscalar@xcite particles between fermions @xmath79 and @xmath80 with @xmath0-couplings to @xmath81 and @xmath82 , gives a @xmath83 potential @xmath84 potentials with @xmath85 are produced by the simultaneous exchange of two massless pseudoscalars with @xmath86 couplings such as axions or goldstone bosons@xcite , but in this case our constraints are not competitive with astrophysical bounds on the first - order process@xcite .
our limits on @xmath87 in table [ tab : power - law constraints ] , together with eq .
[ eq : massless gamma-5 ] , constrain the @xmath0 couplings of massless pseudoscalars to neutrons and protons .
@xmath88 ^ 2 \!\ !
\frac{(g_p^p)^4}{(\hbar c)^2 } \!+ \!\left [ \frac{n}{\mu}\right]^2 \ !
\frac{(g_p^n)^4}{(\hbar c)^2 } \!+\ !
2\left [ \frac{z}{\mu}\right ] \left [ \frac{n}{\mu}\right ] \
! \frac{(g_p^p)^2}{\hbar c}\frac{(g_p^n)^2}{\hbar c } \nonumber \\ & = & \beta_3 \frac{c g}{\hbar^3 } 64 \pi^3 u^4 ( 1~{\rm mm})^2 = 2.56 \times 10^{-10 } \beta_3~. \end{aligned}\ ] ] couplings to electrons can be ignored because of the very small upper limit on such couplings deduced from an electron - spin - dependence experiment ( see ref .
@xcite ) .
we constrained the @xmath0 couplings of pseudoscalars of mass @xmath3 by fitting the ref .
@xcite data in terms of the newtonian potential plus the appropriate generalization@xcite of eq .
[ eq : massless gamma-5 ] , @xmath89 where @xmath90 is a modified bessel function and @xmath91 .
our bounds on @xmath92 are shown in fig .
[ fig : ps constraints ] .
the observed sn1987a neutrino pulse excludes @xmath93@xcite .
stronger couplings are allowed because the pseudoscalars would have been trapped in the star ; we exclude this possibility for @xmath0-coupled pseudoscalars with @xmath94mev .
helioseismology constraints on exotic energy loss processes provide a limit @xmath95@xcite .
kapner et al . , hep - ph/0611184 .
e.g. adelberger , b.r .
heckel and a.e .
nelson , ann .
nucl . part .
sci . * 53 * , 77 ( 2003 ) .
i. antoniadis , s. dimopoulis and g.r .
dvali , nucl .
b * 516 * , 70 ( 1998 ) .
kaplan and m.b .
wise , jhep * 08 * ( 2000 ) .
e. zavattini _ et al . _
( pvlas collaboration ) , phys .
lett . * 96 * 110406 ( 2006 ) .
e. zavattini _ et al .
_ infn - lnl-213 ( 2006 ) .
e. mass and c. rizzo , hep - ph/0610286 .
raffelt , _ stars as laboratories for fundamental physics _ , univ . of chicago press , 1996 .
j. khoury and a. weltman , phys .
lett . * 93 * , 171104 ( 2004 ) .
b. feldman and a.e .
nelson , hep - ph/060307 .
d.f . mota and d.j .
shaw , phys .
* 97 * , 151102 ( 2006 ) .
a. upadhye , s.s .
gubser and j. khoury , hep - ph/0608186 ( 2006 ) .
r. sundrum , phys .
d * 69 * , 044014 ( 2004 ) .
hoyle et al . , phys .
d * 70 * 042004 ( 2004 ) .
r. spero et al .
lett . * 44*,1645 ( 1980 ) ; j.k .
hoskins et al . , phys .
d * * 3**2 , 3084 ( 1985 ) .
j. sucher and g. feinberg , in _ long - range casimir forces _ , eds .
f.s . levin and d.a .
micha ( plenum , new york , 1993 ) pp .
273 - 348 .
f. ferrer and j.a .
grifols , phys .
d * 58 * , 096006 ( 1998 ) .
e. fischbach and d.e .
krause , phys .
lett . * 83 * , 3593 ( 1999 ) .
f. ferrer and m. nowakowski , phys .
d * 59 * , 075009 ( 1999 ) . e.g. adelberger , e. fischbach , d.e . krause and r.d .
newman , phys .
d * 68 * , 062002(2003 ) .
g. raffelt , hep - ph/0611118 . | we use data from our recent search for violations of the gravitational inverse - square law to constrain dilaton , radion and chameleon exchange forces as well as arbitrary vector or scalar yukawa interactions .
we test the interpretation of the pvlas effect and a conjectured `` fat graviton '' scenario and constrain the @xmath0 couplings of pseuodscalar bosons and arbitrary power - law interactions . in a recent letter@xcite
, we reported a sensitive torsion - balance search search for yukawa violations of the gravitational inverse - square law ( isl ) of the form @xmath1 however , space limitations prevented us from discussing some implications of that result and constraining other forms of possible breakdowns of the isl . in this letter
we use the data from ref .
@xcite to obtain upper bounds on several interesting exotic interactions . |
the cabibbo - kobayashi - maskawa ( ckm ) matrix@xcite describes the relationship between weak and mass eigenstates of quarks , assuming that there are three generations . by convention ,
up - type quarks are unmixed such that all the mixing is expressible in terms of the @xmath3 unitary matrix @xmath4 @xmath5 where unprimed states denote mass eigenstates and primed ones denote weak eigenstates .
the unitarity of the ckm matrix is an assumption which must be subjected to experimental verification by independent measurements of the elements .
non - unitarity of the matrix would be a clear signature for physics beyond the standard model , such as a fourth generation or non - universality of the quark couplings .
of particular interest in terms of possible new physics is the element @xmath6 which describes the coupling of the two heaviest quarks .
it is often assumed that @xmath7 although this element has never been determined without assumptions of unitarity .
assuming that there are only three generations and unitarity of the ckm matrix yields : @xmath8 at the 90% confidence level @xcite . relaxing the assumption of three generations but maintaining that of unitarity yields @xmath9 at the 90% confidence level @xcite , while relaxing also that of unitarity leaves @xmath0 unbounded . in this paper
we describe a new method for the determination of @xmath0 from electroweak loop corrections , in particular to the process @xmath1 . from a combined analysis of data from the lep , slc , tevatron , and neutrino scattering experiments , we determine for the first time the value of @xmath0 , without any assumptions about the unitarity of the ckm matrix .
the implications for other quantities , such as the the top mass , higgs mass , and the strong coupling constant , are discussed .
the precision of the electroweak data from the lep and slc experiments is sufficient to be sensitive to weak loop diagrams involving the top quark .
the top quark appears in @xmath11 vacuum polarisation loops , thereby affecting all the @xmath11 partial widths , and in the gim - suppressed vertex diagrams shown at the one - loop level in fig .
[ fig : feynman_diags ] which affect the @xmath11 partial width , @xmath12 , for the process @xmath1 . from fits to the @xmath11 electroweak parameters , the top quark mass has been determined to be @xmath13gev @xcite , in agreement with the direct measurement from the tevatron of @xmath14gev @xcite .
hitherto the theoretical treatments of weak loop corrections to @xmath11 decay processes have assumed that @xmath15 ; we relax this assumption . following the treatment of barbieri , beccaria , ciafaloni , curci , and vicer ( bbccv ) @xcite ,
@xmath12 may be written as : @xmath16 , \ ] ] where @xmath17 is the fermi constant , @xmath18 and @xmath19 are the masses of the @xmath11 and the @xmath20-quark respectively , and @xmath21 includes the effects of radiative corrections to the @xmath11 propagator . @xmath22 and @xmath23 , which are approximately unity , describe the qed and qcd vertex corrections @xcite .
the couplings , @xmath24 and @xmath25 incorporate vertex corrections described by the parameter , @xmath26 , as follows : @xmath27 we have taken the results of the bbccv calculations up to two - loops and included the effects of @xmath6 at the one - loop level , such that @xmath21 is unchanged while @xmath26 is modified by a multiplicative factor of @xmath28 . in the limit @xmath29 , @xmath26 is given by : @xmath30 , \label{eqn : barbtau1}\end{aligned}\ ] ] where @xmath31 .
in the limit @xmath32 , @xmath26 is given by : @xmath33{0.001pt}{2.2em}}}\right .
\right .
311 + 24 \pi^2 + 282 \log{r } + 90 \log^2{r } \nonumber \\ & & - 4 r \left ( 40 + 6 \pi^2 + 15 \log{r } + 18\log^2{r } \right ) \nonumber \\ & & + \frac{3 r^2}{100 } \left(24209 - 6000 \pi^2 - 45420 \log{r } \right .
\nonumber \\ & & \left.\left.\left . -
18000 \log^2{r } \right ) { \phantom{\rule[-0.5em]{0.001pt}{2.2em}}}\right]\right\ } , \label{eqn : barbtau2}\end{aligned}\ ] ] where @xmath34 . in the intermediate region ( @xmath35 ) @xmath26
is described by a polynomial parametrisation of the full bbccv calculation , as a function of @xmath36 , multiplied by a factor of @xmath28 . at the two - loop level some diagrams are of the order @xmath37 .
our treatment of these to order @xmath28 is justified by their relatively small contribution and the current sensitivity of the experimental data , as will be seen below .
the bbccv corrections are incorporated in the zfitter program @xcite which is used by the lep / slc electroweak working groups to derive parameters such as @xmath39 and @xmath40 from @xmath11 data @xcite .
we make modest modifications to zfitter to allow for the effects of @xmath6 described in the previous section . the @xmath11 parameters , from a combined fit to lep / slc data , which we use as input are @xcite : the @xmath11 mass , @xmath41 ; the @xmath11 width , @xmath42 ; the hadronic pole cross - section , @xmath43 ; @xmath44 where @xmath45 is the hadronic partial @xmath11 width and @xmath46 is the leptonic partial width , assuming lepton universality ; and the leptonic pole forward - backward charge asymmetry assuming lepton universality , @xmath47 .
the parameter values , their errors , and their correlation coefficients are shown in table [ tab : lineshape ] .
the parameters pertaining to @xmath20 and @xmath48 quarks which we use are @xcite : @xmath49 ; @xmath50 ; @xmath51 and @xmath52 , the forward - backward charge asymmetries at the @xmath11 pole for @xmath20 and @xmath48 quarks respectively ; and @xmath53 @xmath54 where @xmath55 . the parameter values , their errors , and their correlation coefficients are shown in table [ tab : bcpars ] .
we also use the following parameters which are to a good approximation experimentally uncorrelated : @xmath56 , the average tau polarisation @xcite ; @xmath57 from the tau polarisation forward - backward asymmetry @xcite ; @xmath58 , the left - right asymmetry from sld @xcite ; the qed coupling constant , @xmath59 @xcite ; the strong coupling constant @xmath60 @xcite where the value obtained from the @xmath11 width is not included ; the @xmath61 masses from lep ii @xcite , cdf @xcite , d @xcite , and ua2 @xcite , averaged according to ref .
@xcite ; the top mass from the tevatron @xcite ; and @xmath62 from @xmath63n scattering measurements by charm @xcite , cdhs @xcite , and ccfr @xcite .
the parameter values and their errors are shown in table [ tab : otherpars ] . for given values of @xmath41 , @xmath39 , @xmath40 , @xmath64 , @xmath65 , and @xmath0
our modified version of zfitter provides predictions for all of the parameters shown in tables [ tab : lineshape ] , [ tab : bcpars ] , and [ tab : otherpars ] .
these predictions and the corresponding measured quantities , together with their associated errors and correlation coefficients , are used to construct a chisquare probability @xmath66 .
the minimum of the chisquare is then determined numerically . as a technical cross - check
, we use the same input parameters as those of ward @xcite , set @xmath67 , and successfully reproduce the results for @xmath41 , @xmath39 , @xmath40 , @xmath64 , and @xmath65 .
the results of the fit with @xmath0 free are shown in table [ tab : results ] .
we determine @xmath2 .
this value is consistent with the unitarity prediction of @xmath68 to within approximately one standard deviation . for comparison , table [ tab : results ] also includes the results with @xmath0 fixed to @xmath69 . in both fits ,
the @xmath70 probability is consistent with expectations given the number of degrees of freedom .
the fitted values of @xmath41 and @xmath71 are insensitive to @xmath0 as expected , as shown by the correlation coefficients from the fit with @xmath0 free in table [ tab : correl ] .
the anti - correlations of @xmath0 with @xmath39 and @xmath40 , shown in fig .
[ fig : vtb_fit](a ) and [ fig : vtb_fit](b ) and in table [ tab : correl ] , have only a weak effect on the determinations of @xmath39 and @xmath40 since the tevatron measurement of @xmath39 and the vacuum polarisation contribution to the @xmath11 width constrain @xmath39 and @xmath40 independent of @xmath0 . nonetheless , allowing @xmath0 to float increases the fitted values of @xmath39 by 1.5gev and of @xmath40 by approximately 30gev . allowing @xmath0 to float decreases the fitted value of @xmath64 by approximately 0.7 standard deviations , due to the fairly strong correlation of these two quantities , as shown in fig .
[ fig : vtb_fit](c ) and in table [ tab : correl ] . to assess the future sensitivity of this technique for determining @xmath0 we reduce the error by a factor of two on each of the input parameters in turn , without changing the errors on the other parameters .
the only parameters which cause @xmath72 to change by a relative amount of more than 10% , from the original value of @xmath73 , are @xmath42 , @xmath74 , and @xmath60 , which yield uncertainties of @xmath75 22.5% , 19.9% , and 22.4% respectively .
factors of somewhat less than two may be expected from the final analyses of the lep i and slc data samples .
we estimate that the error on @xmath0 from the final @xmath11 samples of the lep and slc experiments will be approximately 20% .
recently cdf presented preliminary results of an analysis of their @xmath76 event samples .
however , they necessarily assumed that there are only three generations and that @xmath77 , as required by unitarity , to extract @xmath78 @xcite .
ultimately , the measurement of the single top quark production rate at hadron colliders should be sensitive to @xmath0 without requiring such assumptions of unitarity .
the estimated sensitivity at the end of the tevatron run ii for @xmath79 is 12%19% , depending on the uncertainty of the gluon structure functions @xcite .
we describe a new technique for the determination of the ckm matrix element @xmath0 using loop corrections to electroweak processes , without using any unitarity constraints . from a combined analysis of data from the lep , slc , tevatron and neutrino scattering experiments we determine @xmath2 where the error includes the experimental errors and uncertainties on the top mass , the higgs mass , and the strong coupling constant
. the wider implications of this measurement are discussed elsewhere @xcite .
we would like to thank joachim mnich for many valuable discussions , the national science foundation for financial support , and the department of physics , universidad nacional de la plata for their generous hospitality while this work was being completed .
n. cabibbo , phys .
* 10 * , 531 ( 1963 ) , m. kobayashi and t. maskawa , prog .
. phys . * 49 * , 652 ( 1973 ) .
, phys . rev .
* d54 * , 1 ( 1996 ) .
d. ward , invited talk at the _ international europhysics conference on high energy physics _ , jerusalem , 1997 , ( unpublished ) .
r. raja , in _ proceedings of the xxxiind rencontres de moriond , moriond , 1997 _ edited by j. trn thanh vn , editions frontires , gif - sur - yvette , ( 1997 ) . .
d. bardin _
et al . _ , in cern report * 95 - 03 * , edited by d. bardin , w. hollik and g. passarino , ( 1995 ) , ( unpublished ) .
d. bardin _
et al . _ , cern - th * 6443 - 92 * , ( 1992 ) , ( unpublished ) ; * hep - ph/9412201 * , ( 1994 ) , ( unpublished ) .
s. eidelmann and f. jegerlehner , z. phys .
* c67 * , 585 ( 1985 ) . . .
* b276 * , 365 ( 1992 ) . , hep - ex/*9712016*. to appear in the _ proceedings of the xviith international conference on physics in collision _ ( 1997 ) , edited by h. heath , world scientific , singapore . . .
, fnal - pub-97 * 001-e * , ( 1997 ) , ( unpublished ) .
heinson , in _ proceedings of the 2nd international conference on b physics and cp violation , honolulu , march , 1997 _ , * hep - ex/9707026 * ( unpublished ) . , phys . rev .
* d56 * , 3114 ( 1997 ) .
, hep - ph/*97xxxxx * ( unpublished ) . | we present a new method for the determination of the cabibbo - kobayashi - maskawa quark mixing matrix element @xmath0 from electroweak loop corrections , in particular those affecting the process @xmath1 . from a combined analysis of results from the lep , slc , tevatron , and neutrino scattering experiments we determine @xmath2 .
this is the first determination of @xmath0 which is independent of unitarity assumptions . |
detecting electron counting statistics has become a major experimental challenge in mesoscopic physics .
first attempts to measure non - gaussian effects in current noise have revealed that the detection problem is quite subtle . in particular , the experiment@xcite found that the third current cumulant was not described by the simple theoretical prediction,@xcite but was masked by the influence of the measurement circuit causing an additional `` cascade '' correction.@xcite recent experiments demonstrated a measurement of the third current cumulant without cascade corrections,@xcite and the detection of individual electron counting statistics.@xcite stringent bandwidth requirements in measuring the third cumulant suggested that further experimental advances would require a new approach . a conceptually different way to measure rare current fluctuations is with a threshold detector,@xcite the basic idea of which is analogous to a pole vault : a detection event occurs when the measured system variable exceeds a given value .
a natural candidate for such a detector is a metastable system operating on an activation principle.@xcite by measuring the rate of switching out of the metastable state , information about the statistical properties of the noise driving the system may be extracted . a threshold detector using an on - chip conductor which contains a region of negative differential resistance@xcite was proposed by the authors and shown to be capable of measuring large deviations of current .
tobiska and nazarov proposed a josephson junction ( jj ) threshold detector,@xcite the simplest variant of which ( see fig . [ circuit ] ) operates essentially in a gaussian regime.@xcite the third cumulant contribution is small@xcite and may be extracted using the asymmetry of the switching rate with respect to bias current , as has been demonstrated in recent experiments.@xcite = 6.5 cm [ circuit ] in this letter we solve kramers problem@xcite of noise - activated escape from a metastable state beyond the gaussian noise approximation and investigate how the measurement circuit affects threshold detection . starting with general hamiltonian - langevin equations which includes deterministic dynamics , dissipation , and fluctuations ,
we represent the solution as a stochastic path integral of hamiltonian form@xcite by doubling the number of degrees of freedom . in the weak damping case ,
the dynamics is dominated by energy diffusion , which we account for by a change of variables , enabling an effectively two - dimensional representation .
we calculate the escape rate via an instanton calculation , and obtain a formal solution of kramers problem.@xcite applying these general results to a jj threshold detector , we account for the influence of the measurement circuit and find that the cascade corrections are a consequence of the non - equilibrium character of the noise.@xcite
in order to reformulate and solve kramers problem beyond gaussian noise , we introduce a theoretical framework that relies on a separation of time scales : slow motion of a deterministic system on a time scale @xmath0 is affected by quickly fluctuating noise sources with correlation time @xmath1 .
quite generally , this situation can be described by hamiltonian - langevin ( hl ) equations .
for instance , in a two - dimensional phase space @xmath2 the equations of motion are @xmath3 where @xmath4 is the hamiltonian that generates deterministic motion , and @xmath5 and @xmath6 are white noise sources . on the intermediate time scale @xmath7 ,
such that @xmath8 , the sources are markovian and fully characterized by the generating function @xmath9 of their cumulants ( irreducible moments ) : @xmath10 .
while langevin equations may be solved by standard methods in stochastic physics,@xcite the methods fail when applied to higher - order cumulants .
the reason is that slow variations of @xmath11 and @xmath12 affect the noise sources and lead to _
cascade corrections_,@xcite which can not be obtained from linearized differential equations ( [ hle ] ) . following the steps of refs .
[ ] , we represent the slow evolution of the system by the stochastic path integral ( spi ) @xmath13 , where the action @xmath14 is given in the explicitly canonically invariant form as @xmath15 . \label{action}\ ] ] here @xmath16 and @xmath17 are the sets of physical and canonically conjugated `` counting '' variables , respectively , and @xmath18 denotes the poisson bracket with respect to @xmath11 and @xmath12 . by fixing @xmath19 in the final state of ( [ action ] )
we obtain the probability distribution @xmath20 , while fixing the final @xmath21 variables turns the spi into the moment generating function @xmath22 .
the large parameter @xmath23 allows the saddle - point evaluation of the spi and thus requires solving hamilton s equations of motion in the extended space.@xcite there always exists a trivial solution @xmath24 and @xmath25 , where @xmath26 , that describes the `` average '' dynamics in physical space with a null action @xmath27 , giving the proper normalization of the distribution @xmath28 . in the context of the generalized kramers problem , one has to find a non - trivial _ instanton _ solution @xmath29 of eq .
( [ action ] ) , for which the spi gives a rate of the noise activated escape from a localized state , @xmath30 .
as an important example , let us consider quasi - periodic motion with the period @xmath31 .
we change variables to energy @xmath32 and `` time '' @xmath33,@xcite accompanied by a new set of counting variables @xmath34 , which are defined via @xmath35 and @xmath36 . using @xmath37 and @xmath38
, we obtain @xmath39 .
fluctuations of the @xmath40 variable increase the action without leading to escape , therefore they can be neglected by choosing @xmath41 . in the weak damping regime , one can set @xmath42 , so that @xmath43 and @xmath44 .
the action for the energy diffusion then reads , @xmath45 .
\label{diffusion}\ ] ] the following intuitive argument supports this result .
the energy balance equation @xmath46 follows directly from the hl equations ( [ hle ] ) and takes the form of a langevin equation .
if @xmath31 , then variables @xmath47 and @xmath48 change slowly and can be considered as being `` effective charges '' , which after following [ ] leads to eq .
( [ diffusion ] ) . to leading order in weak damping
we can replace the generator @xmath49 in eq .
( [ diffusion ] ) with its average over the oscillation period @xmath50 , evaluated for fixed @xmath51 and @xmath52 .
corrections in damping may be found by taking into account slow energy dissipation @xmath53 and @xmath54 , while averaging over the period @xmath0 .
we are interested in the instanton solution @xmath55 with @xmath56 in the initial and final state.@xcite since the `` hamiltonian '' @xmath57 is an integral of motion , we obtain an important result for the escape rate , @xmath58 which formally solves kramers problem for arbitrary markovian noise in the weak damping limit .
below we apply the theory outlined here to the stochastic dynamics of a josephson threshold detector .
the circuit in fig . 1 shows the essential part of the detector comprised of the jj with josephson energy @xmath59 , and the capacitor , @xmath60 .
the circuit is current - biased with @xmath61 through the macroscopic conductor and by the system current @xmath62 , which is to be measured . according to kirchhoff s law
, the total bias current @xmath63 is equal to the sum of the josephson current @xmath64 where @xmath65 , and the displacement current @xmath66 .
this leads to the equation of motion for the superconducting phase @xmath67 , @xmath68 where we used the relation @xmath69 . in order to simplify the following analysis and concentrate on our main message
, we make a number of assumptions , most of which will be relaxed later .
first , we consider an ohmic system and bias resistor , so that @xmath70 and @xmath71 , where @xmath72 , @xmath73 are the system and bias conductances , and the constant currents @xmath74 , @xmath75 are just tunable parameters .
the bias resistor , being a macroscopic system , creates gaussian nyquist noise @xmath76 .
we further assume a high - impedance circuit , so that the back flow part of the bias current @xmath77 and the nyquist noise may be neglected , @xmath78 , which we refer to as the _ ideal detection _ scheme .
then eq .
( [ motion ] ) can be rewritten as a set of hl equations ( [ hle ] ) for the phase variable @xmath67 and canonically conjugated momentum @xmath79 ( where @xmath80 is the total charge on the capacitor ) @xmath81 with `` mass '' @xmath82 . equations ( [ hle - phase ] ) describe the motion of a `` particle '' in the tilted periodic potential @xmath83 stimulated by the dissipative part @xmath84 of the system s current . for later convenience
, we define the normalized total bias current as @xmath85 .
dissipation leads to relaxation of the jj into one of its metastable states . in the supercurrent state ,
the phase is localized in one of the potential wells , so that @xmath86 . in the dissipative state , the phase drifts along the bias which generates the non - zero voltage drop @xmath87 . in the remaining part of the paper we investigate noise
activated escape from the localized state .
here the system conductance is small , @xmath88 , so the phase oscillates with the plasma frequency @xmath89 , where @xmath90 .
the energy relaxes slowly with the rate @xmath91 to the local potential minimum .
we further assume the separation of time scales , @xmath92 , so that the noise source @xmath62 is markovian . comparing ( [ hle - phase ] ) and ( [ hle ] ) ,
we identify @xmath93 , @xmath94 and @xmath95 , so the equations for the escape rate and `` instanton line '' read @xmath96 where @xmath49 is the generator of the cumulants of the dissipative part of the system current , @xmath84 , and @xmath97 plays the role of an effective charge .
we first consider the system in thermal equilibrium with gaussian nyquist noise , @xmath98 , as a checkpoint of the theory .
the cumulant generator acquires the simple form @xmath99 , where the first term comes from the linear response of the system current @xmath100 to the potential @xmath69 , while the second term is the noise contribution . averaging @xmath49 over the period of oscillations , we observe that the term @xmath101 cancels , so that @xmath102 . using eq .
( [ wdr - instanton ] ) we obtain kramers well - known formula for the rate of thermally activated escape @xmath103 where the potential threshold @xmath104 is a function of normalized bias @xmath105 . for the potential ( [ potential ] ) one obtains @xmath106 , see fig .
[ parameters ] . limiting values are @xmath107 for @xmath108 , while as @xmath109 , @xmath110 . [ parameters ] the above example
shows that the argument of @xmath49 is small .
indeed , we estimate @xmath111 , so @xmath112 due to the separation of time scales . therefore , for any system away from equilibrium we can expand @xmath113 in eq .
( [ wdr - instanton ] ) , average the result over the period of oscillations , and invert the series for @xmath114 . to second order in @xmath114
we again obtain kramers formula ( [ kramers ] ) with the temperature @xmath115 replaced with the effective temperature @xmath116 of the non - equilibrium noise . to find the third cumulant correction to the kramers formula , we expand @xmath117 in @xmath114 and the voltage @xmath87 , and collect all terms to next order in small parameter @xmath118 .
two terms which contain @xmath119 and @xmath120 come with the factor @xmath121.@xcite this factor vanishes to leading order in damping , because for fixed energy @xmath122 is an odd function of time over the oscillation period .
therefore , we evaluate the average @xmath121 by taking into account slow variation of the energy .
this is accomplished by taking functional variations of the action to find @xmath123 and @xmath124 .
we skip a number of straightforward steps ( see appendix ) and present the result for the non - equilibrium escape rate @xmath125 [ wdr ] where the dimensionless factor @xmath126 is a characteristic of the detector . appearing in the integral ( [ d1 ] )
is one of the turning points of the oscillating phase , @xmath127 , defined by the solution of @xmath128 that is nearest the top of the potential .
for the potential ( [ potential ] ) @xmath129 and for strong bias @xmath130 we obtain @xmath131 .
the parameter @xmath132 , evaluated numerically , is shown in fig .
[ parameters ] . the total third current cumulant @xmath133 is taken at @xmath134 and contains a correction originating from the slowly varying second current cumulant .
this contribution is analogous to the cascade correction@xcite directly observed in third cumulant.@xcite the correction is generally not small and may even change the sign of the total cumulant .
for instance , for systems far from equilibrium @xmath135 , where @xmath136 are the dimensionless fano factors , we obtain @xmath137 .
in this regime the conductance is large , @xmath138 , and the dynamics is entirely due to slow phase relaxation with the rate @xmath139 .
therefore , we can set @xmath140 and @xmath141 , and neglect the first term in the action ( [ action ] ) , so that the action reads : @xmath142 $ ] , where again @xmath49 is the generator of the cumulants of the dissipative part of the system current , @xmath84 .
the following analysis is analogous to that of the weak damping regime .
we first expand @xmath49 to second order in @xmath143 and find from the equations of motion that @xmath144 and @xmath145 . substituting these results back to the action we immediately obtain kramers formula ( [ kramers ] ) with @xmath115 replaced with @xmath146 .
next we observe that the argument of @xmath49 is small , namely @xmath147 due to the separation of time scales requirement .
we collect all the terms to third order in this parameter and evaluate them perturbatively using the above results for @xmath143 and @xmath122 .
we finally obtain @xmath148 [ sdr ] where the total cumulant @xmath149 is given by eq .
( [ cumulant1 ] ) , and @xmath150 is a dimensionless detector property . for the potential ( [ potential ] ) it is given by @xmath151 , see fig .
2 . limits are @xmath152 , and as @xmath109 , @xmath153 .
we now remark on the application of our results to the detection of non - gaussian fluctuations .
it is evident from fig .
[ parameters ] that @xmath154 as @xmath130 .
therefore , in the strong bias regime the third cumulant contribution in ( [ wdr ] ) and ( [ sdr ] ) is suppressed compared to the kramers term @xmath155 . but even for a relatively weak bias , when @xmath156 , non - gaussian effects are small .
indeed , it is easy to estimate the correction as @xmath157 , where the jj quality factor @xmath158 in the weak damping regime ( wd ) , and the ratio @xmath159 due to the separation of time scales .
similarly , in the strong damping regime ( sd ) , @xmath160 , the correction is of order @xmath161 , where @xmath162 due to the separation of time scales . since @xmath163 itself can not be too large for the escape to be detected , in experiments one should try to saturate the above inequalities and use the asymmetry of the third cumulant ( [ cumulant1 ] ) as a function of the current bias @xmath164 .
next , we briefly discuss nonideal detection and nonlinear effects .
the finite conductance of the bias resistor @xmath73 contributes to the total conductance of the circuit @xmath165 and the nyquist noise @xmath166 adds to the total noise power .
therefore , in eqs .
( [ wdr]-[sdr ] ) one has to replace @xmath72 with @xmath167 and the effective temperature with @xmath168 .
the nonlinearity of the system current leads to the correction @xmath169 to @xmath49 , and thereby to an additional contribution to the total third cumulant , @xmath170 , in both transport regimes considered above .
quite remarkably , this total third cumulant is related via @xmath171 to the instantaneous fluctuations of the voltage @xmath87.@xcite we finally note that in the experiment of ref . [ ] the circuit corrections have not been observed , which we explain by a very low impedance of the circuit , @xmath172 , operating in the weak damping regime with @xmath173 . indeed , in this case @xmath174 ( assuming @xmath175 ) , therefore the circuit corrections are suppressed by a small factor @xmath176 . on the other hand , the first term in eq .
( [ rate1 ] ) is @xmath177 and the second term can be estimated as @xmath178 , where @xmath179 in this case .
therefore the relative contribution of the third cumulant to the total action increases by the factor @xmath180 , which makes it favorable to use a low - impedance circuit . in conclusion
, we have formally solved kramers problem of noise - activated escape from a metastable state for markovian non - equilibrium noise beyond the gaussian approximation .
we have applied this result to the josephson junction threshold detector and evaluated the third current cumulant contribution to the escape rate .
the back action of the electrical circuit on the measured noise is entirely a non - equilibrium effect , which leads to `` cascade '' corrections for the third cumulant
. it would be interesting to apply our theory to the dynamics of the josephson bifurcation amplifier.@xcite in this appendix , we give the details of the derivation of eq .
( [ wdr ] ) . starting with eq .
( [ wdr - instanton ] ) , we expand the generating function in the counting variable to third order , @xmath181 taking the assumptions in the text of an ohmic system conductor , we replace @xmath182 .
the noise produced by the system conductor depends on the dynamically changing voltage drop @xmath183 ( see fig . [ circuit ] ) , so we also expand the second current cumulant in the voltage @xmath87 , namely @xmath184 , where we recall that @xmath185 is the voltage across the system , and the coefficients above are evaluated at @xmath134 . using the josephson relation @xmath186 , and averaging ( [ a1 ] ) over the oscillation period at fixed energy , we now need to find the instanton line and solve the equation @xmath187 for @xmath188 . as argued in the text ,
the instanton solution @xmath188 is small for the jj , so the series may be inverted to find to third order in @xmath189 , @xmath190 inserting this instanton solution into the action , we obtain for the activation rate @xmath191 however , to leading order in damping , the average @xmath192 vanishes because @xmath193 is an odd function of time over the interval of one period .
therefore , we must evaluate this average to next order in the variation of energy over one period .
recalling that the action is @xmath194 $ ] , the time dependence of the energy @xmath52 may be found by taking the functional derivative @xmath195 to find @xmath196 . inserting the instanton solution @xmath197
, we obtain @xmath198 so the energy is increasing . at a given value of @xmath67 , the linear correction to @xmath199 due to energy dissipation is given by @xmath200 .
therefore , we use @xmath201 to write @xmath202 where the integral starts from the turning point nearest to the top of the barrier , @xmath203 .
this is done because the escape of the jj into the running state happens at the top of the barrier , so the last part of the trajectory before escape must be fully accounted for . integrating by parts , @xmath204 , we keep only the change in energy , and neglect the change in the turning point to this order . using the result ( [ edot ] ) ,
these approximations imply that @xmath205 . collecting results , we find @xmath206 recalling the definition @xmath207 , and inserting ( [ third1 ] ) into ( [ result ] )
, we recover our main result ( [ wdr ] ) .
the prefactor in @xmath208 remains unknown , because it depends on non - markovian behavior at short times of order @xmath209 . for a discussion of the prefactor in a jj context ,
see , e.g. , m. bttiker , e. p. harris , and r. landauer , phys .
b * 28 * , 1268 ( 1983 ) . | we generalize the stochastic path integral formalism by considering hamiltonian dynamics in the presence of general markovian noise .
kramers solution of the activation rate for escape over a barrier is generalized for non - gaussian driving noise in both the overdamped and underdamped limit .
we apply our general results to a josephson junction detector measuring the electron counting statistics of a mesoscopic conductor .
activation rate dependence on the third current cumulant includes an additional term originating from the back - action of the measurement circuit . |
in the lattice calculation of the @xmath6 meson @xmath6-parameters @xcite , the use of the effective theory for heavy quark is essential to obtain results with controlled systematic errors .
the static approximation in which the heavy quark mass is sent to infinity has been used by many authors @xcite , and the next step is to incorporate the correction from finite @xmath7 quark mass . in this direction , we performed an exploratory lattice calculation using the nrqcd action for heavy quark , in which we found that the @xmath8 correction could be sizable for @xmath1 @xcite and for @xmath2 @xcite .
a drawback in these calculations was , however , that the perturbative renormalization of the four - quark operators were not available for the nrqcd action and we used the one - loop coefficients for the static action . as a result our results contained large systematic error of @xmath5 , which is about 10% if evaluated with an order counting at @xmath9 = 5.9 . in this paper , we present a perturbative calculations of renormalization constants of the heavy - light four quark operators with the @xmath3 lattice nrqcd action .
the results connect the missing link in our previous calculations , and we present a reanalysis of @xmath1 and @xmath2 . a full description of our results is published in @xcite .
the @xmath6 parameters in the @xmath10 are defined through @xmath11 @xmath12 where a subscript @xmath13 denotes light quark flavor , i.e. @xmath14 or @xmath15 .
@xmath16 is usually called @xmath1 , but we use a notation @xmath16 in this paper to remind that it parameterizes a matrix element of the operator @xmath17 . in the above equations , the operators @xmath18 and @xmath19 are defined in the continuum theory with the @xmath20 scheme .
their dirac structure is @xmath21 where @xmath22 .
the totally anti - commuting convention for @xmath23 is assumed in the dimensional regularization .
the matching of these continuum operators is done against the lattice nrqcd .
namely we consider the @xmath3 nrqcd action for heavy quark and the @xmath4-improved wilson fermion for light quark . for gluons the standard plaquette action is used .
for @xmath24 and @xmath25 the following heavy - light four - quark operators appear in the matching relation : @xmath26 \left [ \bar{b}\gamma_\mu p_l q\right],\nonumber\\ & & \hspace*{-2em } o_r = \left [ \bar{b}\gamma_\mu p_r q\right ] \left [ \bar{b}\gamma_\mu p_r q\right],\nonumber\\ & & \hspace*{-2em } o_s = \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_l q\right],\nonumber\\ & & \hspace*{-2em } o_n = 2 \left [ \bar{b}\gamma_\mu p_l q\right ] \left [ \bar{b}\gamma_\mu p_r q\right ] + 4 \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_r q\right],\nonumber\\ & & \hspace*{-2em } o_m = 2 \left [ \bar{b}\gamma_\mu p_l q\right ] \left [ \bar{b}\gamma_\mu p_r q\right ] - 4 \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_r q\right],\nonumber\\ & & \hspace*{-2em } o_p = 2 \left [ \bar{b}\gamma_\mu p_l q\right ] \left [ \bar{b}\gamma_\mu p_r q\right ] + 12 \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_r q\right],\nonumber\\ & & \hspace*{-2em } o_t = ( 2+n_c ) \left [ \bar{b}\gamma_\mu p_l q\right ] \left [ \bar{b}\gamma_\mu p_r q\right ] \nonumber \\ & & \hspace*{2em } -2(3 n^2_c - 2 n_c - 4 ) \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_r q\right],\end{aligned}\ ] ] where @xmath27 . the heavy quark field @xmath7 has four dirac components , and it is obtained from the two - component spinor non - relativistic quark ( anti - quark ) , @xmath28 ( @xmath29 ) , through the foldy - wouthuysen - tani transformation , @xmath30 \left ( \begin{array}{c } q \\ \chi^{\dag } \end{array } \right).\ ] ] in the static limit @xmath31 , where algebra is greatly reduced
, we consider the @xmath4-improvement of the lattice four - quark operators through @xmath32 . in
doing that we also define the following dimension seven operators : @xmath33 \left [ \bar{b}\gamma_\mu p_l ( a \vec{d}\cdot\vec{\gamma } ) q\right],\nonumber\\ o_{sd } & = & \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_l ( a \vec{d}\cdot\vec{\gamma } ) q\right],\nonumber\\ o_{nd } & = & 2 \left [ \bar{b}\gamma_\mu p_l q\right ] \left [ \bar{b}\gamma_\mu p_r ( a \vec{d}\cdot\vec{\gamma } ) q\right]\nonumber \\ & & + 4 \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_r ( a \vec{d}\cdot\vec{\gamma})q\right ] , \nonumber\\ o_{pd } & = & 2 \left [ \bar{b}\gamma_\mu p_l q\right ] \left [ \bar{b}\gamma_\mu p_r ( a \vec{d}\cdot\vec{\gamma } ) q\right]\nonumber \\ & & + 12 \left [ \bar{b } p_l q\right ] \left [ \bar{b } p_r ( a \vec{d}\cdot\vec{\gamma})q\right].\end{aligned}\ ] ]
at one - loop level , the matching of the continuum operator @xmath34 is given by @xmath35 o^{\rm lat}_{x}(1/a ) \nonumber \\ & & + \sum_{y={\rm dim .
} 6 } \frac{\alpha_s}{4\pi}{\zeta_{x , y}(\mu;1/a ) } o^{\rm lat}_{y}(1/a ) \nonumber \\ & & + \sum_{z={\rm dim . } 7 } \frac{\alpha_s}{4\pi}\zeta_{x , z } o^{\rm lat}_{z}(1/a),\end{aligned}\ ] ] where the summations with @xmath36 and @xmath37 run over dimension six and seven operators respectively .
@xmath38 is the renormalization scale in the @xmath39 scheme and @xmath40 is lattice spacing . the coefficients @xmath41 are determined so that the on - shell scattering amplitudes in the continuum and lattice theories agree with each other . for the external state ,
zero spatial momentum heavy and light quark suffice to determine the coefficients for dimension six operators .
a derivative in terms of spatial momenta have to be considered to obtain the dimension seven coefficients .
the matching relation for @xmath24 is @xmath42 o_{l}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4\pi } \zeta_{l , s}(1/a)\ o_{s}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4\pi } \zeta_{l , n}(1/a)\ o_{n}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4\pi } \zeta_{l , m}(1/a)\ o_{m}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4\pi } \zeta_{l , r}(1/a)\ o_{r}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4\pi } \zeta_{l , ld } o_{ld}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4\pi } \zeta_{l , nd}\ o_{nd}^{\rm lat}(1/a ) .
\label{eqol}\end{aligned}\ ] ] the coefficient @xmath43 of the leading operator @xmath44 contains logarithmic terms @xmath45 , while others do not have @xmath38 dependence .
the operator @xmath46 does not exist in the matching of the static operator , but it becomes necessary for nrqcd .
figure [ fig : zeta_lx ] shows the heavy quark mass dependence of @xmath47 for the dimension six operators . filled symbols at @xmath48 are calculated for the static action .
we find a large slope in @xmath49 for @xmath50 as shown in the top panel of figure [ fig : zeta_lx ] .
most of the mass dependence comes from factorized diagrams , i.e. the four - quark operator can be split into two axial - vector currents and the gluon propagator does not connect these two . as a result , the coefficient @xmath50 is much reduced if we subtract the contributions of two axial vector current , as plotted in the lower panel ( @xmath51 ) .
this combination will appear in the analysis of the @xmath6 parameters , when we consider the ratio @xmath52 .
the coefficients for the dimension seven operators are obtained in the static limit as @xmath53 = @xmath5417.20 and @xmath55 = @xmath549.20 @xcite .
the matching of @xmath56 is given by @xmath57 o_s^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4 \pi } \zeta_{s , l}(\mu;1/a ) o_l^{\rm lat}(1/a ) \nonumber\\ & & + \frac{\alpha_s}{4 \pi } \zeta_{s , p}(1/a ) o_p^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4 \pi } \zeta_{s , r}(1/a ) o_r^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4 \pi } \zeta_{s , t}(1/a ) o_t^{\rm lat}(1/a ) \nonumber\\ & & + \frac{\alpha_s}{4 \pi } \zeta_{s , sd } o_{sd}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4 \pi } \zeta_{s , ld } o_{ld}^{\rm lat}(1/a ) \nonumber \\ & & + \frac{\alpha_s}{4 \pi } \zeta_{s , pd } o_{pd}^{\rm lat}(1/a ) , \label{eqos}\end{aligned}\ ] ] where @xmath58 is new for the nrqcd action .
the coefficients @xmath59 are plotted in figure [ fig : zeta_sx ] . like @xmath50 in the matching of @xmath17 , the leading operator
@xmath56 has large coefficient , which is reduced by subtracting the factorized contribution , as shown in the upper panel .
the coefficients for the dimension seven operators are @xmath60 = @xmath546.88 , @xmath61 = 2.58 and @xmath62 = 1.15 in the static limit .
with these perturbative coefficients , we reanalyze our previous lattice calculation for the @xmath6 parameters , @xmath63 and @xmath64 @xcite . the @xmath6 parameter @xmath63 is calculated through @xmath65 where @xmath66 runs over @xmath67 , @xmath68 , @xmath69 , @xmath70 and @xmath71 .
the renormalization constants are @xmath72 and the lattice @xmath6 parameters @xmath73 were measured in the numerical simulations through @xmath74 in our previous paper @xcite . since the dimension
seven operators are not included in the calculation , we could not remove the error of @xmath75 .
figure [ fig : z_l , x*b_x ] shows the individual contributions @xmath76 for the @xmath6 parameter @xmath63 .
the @xmath77-scheme coupling constant @xmath78 is used with @xmath79 .
although there is a sizable mass dependence in @xmath80 as we can see from the plot of coefficients ( figure [ fig : zeta_lx ] ) , it is canceled by the @xmath8 corrections in the matrix elements on the lattice @xmath81 , and there is little mass dependence in the products shown in figure [ fig : z_l , x*b_x ] .
our result for @xmath82 is plotted in figure [ fig : b_l ] as a function of @xmath83 with filled circles .
the statistical error shown by solid error bars is very small except for the heaviest heavy quark , while the systematic error given by dotted error bars is more significant .
we estimate the systematic errors with an order counting , in which the systematic errors of @xmath84 , @xmath75 , @xmath85 , @xmath86 , and @xmath87 are considered and added in quadrature . for comparison
, we also show a result with one - loop coefficients calculated for the static heavy quark by open circles , for which a large systematic error of @xmath88 is added .
the effect of the nrqcd renormalization constants is sizable , but still it is within the estimated systematic error .
the numerical results for @xmath16 at physical @xmath6 meson mass are @xmath89 where the first error is statistical and the second is systematic one ( @xmath90 ) .
the su(3 ) breaking effect is @xmath91 where only the statistical error is quoted . using eq .
( [ eqos ] ) , @xmath64 is obtained as @xmath92 where @xmath66 runs over @xmath68 , @xmath67 , @xmath71 , @xmath93 and @xmath94 . in the lhs
, we define a ratio of heavy - light matrix elements @xmath95 the renormalization constants are @xmath96 for @xmath2 we define the lattice @xmath6 parameters as @xmath97 the mass dependence of @xmath98 is plotted in figure [ fig : b_l ] ( filled squares ) .
open squares are our previous results calculated with the static renormalization constants .
the mass dependence is less significant in the results with the correct ( nrqcd ) renormalization constants .
the numerical results are @xmath99 and @xmath100 for the su(3 ) breaking ratio .
s.h . and t.o .
are supported in part by the grants - in - aid of the ministry of education ( nos . 11740162 , 12640279 ) .
and n.y . are supported by the jsps research fellowship .
a. k. ewing _
et al . _ ,
d * 54 * ( 1996 ) 3526 ; v. gimnez and g. martinelli , phys .
b398 * ( 1997 ) 135 ; j. christensen , t. draper , and c. mcneile , phys .
d * 56 * ( 1997 ) 6993 ; v. gimnez and j. reyes , nucl .
b545 * ( 1999 ) 576 , hep - lat/0009007 , these proceedings . | we present a perturbative calculation of the renormalization constants for the @xmath0 four - quark operators , which is needed to evaluate the bag parameters @xmath1 and @xmath2 in lattice qcd .
the one - loop coefficients are calculated for the @xmath3 nrqcd heavy quark action and the @xmath4-improved wilson action for light quark .
using these coefficients , which remove @xmath5 error , we reanalyze our previous calculation of the bag parameters . |
the current cosmological paradigm postulates that large galaxies formed from the hierarchical aggregation of smaller ones ( e.g. , * ? ? ?
* ) . within this context
, a galaxy such as the milky way ( mw ) is expected to have an assembly path where infall , mergers , and interactions contribute , with their different relative importance influenced by the local environment .
cosmological simulations are the most suitable tool to describe this complex history of formation , and to confront the adopted cosmological scenario and baryonic sub - grid models with observations .
sophisticated numerical codes have been developed over the past few years , and they are continuously being updated to improve their representation of different physical processes ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in particular , the treatment of chemical evolution within cosmological codes provides a powerful tool for exploring galaxy formation , and to confront models and observations ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) . in this regard , stellar haloes , such as those of the mw and m31 , are valuable laboratories for confronting expectations from cosmological models , since most of their stars are quite old , and were born in the first epochs of galaxy formation .
their chemical abundance patterns , preserved basically unchanged over their long lifetimes , provide a fossil record of the physical characteristics of the interstellar medium at the time and place of their birth ( e.g. , * ? ? ?
we also acknowledge the need to better develop the sets of questions to be posed to models , so that their full utility for exploration of the expected behaviour of actual galaxies can be realized . in this paper
, we make a first effort in this direction .
recent observations provide evidence that the mw halo system comprises at least two overlapping stellar components , an inner - halo population and an outer - halo population , with different metallicities , kinematics , and spatial density profiles ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
these findings suggest a dual history of halo formation , with contributions from stars accreted from lower - mass sub - galactic systems and those formed .
important additional information on the nature of the halo of the mw has begun to emerge from consideration of the chemical abundances of the light and heavy elements .
for example , @xcite have shown that clear differences exist between the frequencies of the so - called carbon - enhanced metal - poor ( cemp ) stars that they kinematically associate with the inner- and outer - halo populations .
they also present evidence for the existence of a correlation between the frequency of cemp stars and distance from the galactic plane , a result which is difficult to understand in the absence of ( at least ) a dual halo .
@xcite has argued that kinematically separated members of the inner- and outer - halo population exhibit a mean [ mg / fe ] ratio for the outer halo that is lower than that of the inner halo , by about 0.1 dex .
he also demonstrated that the star - to - star scatter in [ ni / fe ] and [ ba / fe ] for stars associated with the inner halo is consistently smaller than that of the outer halo .
the existence of different stellar populations located within the inner - halo region of the mw has recently emerged from studies of the @xmath2-elements ( e.g. , * ? ? ?
* ; * ? ? ?
clearly separable correlations have been shown to exist between [ @xmath2/fe ] and [ fe / h ] ( ` high-@xmath2 ' and ` low-@xmath2 ' sequences ) , as well as with kinematics .
most recently , @xcite have extended these differences to include ages .
state - of - the - art observations are also providing information on the stellar haloes of nearby galaxies .
the nature of the stellar halo of m31 has only recently become feasible to explore in detail , due to large - scale surveys enabled by modern detectors as well as deep hubble space telescope observations(e.g .
* ; * ? ? ?
sarajedini et al .
presented evidence that the m31 spheroid possesses a bimodal metallicity distribution , similar in many ways to that suggested by @xcite to apply to the mw .
in addition , m32 has been reported to exhibit different spheroid populations , with the outer one being @xmath3 gyr older and @xmath4 dex more metal poor than its inner halo @xcite . from a theoretical point of view
, cosmological simulations of galaxy formation within the concordance @xmath5-cdm model have also made progress in the description of the formation history of stellar haloes .
@xcite showed a dual history of formation for stellar haloes , with contributions from both accreted sub - galactic systems and star formation , the latter principally located in the central regions of the haloes .
more recently , similar findings have been reported by @xcite , @xcite and @xcite , through the analysis of stellar haloes in simulated galaxies using a variety of numerical codes and sub - grid physics models .
considering the different approaches used for the baryonic sub - grid models among these cosmological codes , it is noteworthy that they all agree on suggesting a dual formation history for the stellar haloes , in the sense that they are formed by the combination of stars born and in accreted sub - galactic systems . this simple way of describing a complex assembly process , which depends strongly on the galaxy evolutionary path , has been successful at illustrating the global picture . to move forward , in this work we aim at studying in more detail the origin of each different stellar populations that form
the diffuse haloes in a @xmath5-cdm scenario .
the results discussed in this paper are related to those reported by ( * ? ? ?
* hereafter , paper i ) who analysed the chemical properties of galaxies formed in eight mw mass - size systems of the aquarius project , with different assembly paths within the @xmath5-cdm scenario @xcite .
paper i shows that , although the central regions of these simulated galaxies and their stellar haloes are dominated by very old stars , their chemical properties differ as a consequence of their individual formation histories .
the former were dominated by star formation , the latter were assembled mainly from small accreted sub - galactic systems .
although the mass scales and time scales involved resulted in different chemical patterns at @xmath6 , these simulations successfully reproduced the global chemo - dynamical trends observed in the mw .
following paper i , we adopt binding energy criteria defined to dynamically separate two stellar halo populations without any assumptions on the chemical abundances or kinematics of the stars . hereafter , to clarify the description of our results , the most gravitationally bound stars , dynamically associated with the so - called inner haloes , will be referred to as inner - halo populations ( ihps ) , while the less bound stars , associated with the so - called outer haloes , will be referred to as outer - halo populations ( ohps ) .
note that , as the separation of these stellar components is performed on the basis of their binding energy content , the ihps and ohps can be spatially superposed .
we thus also distinguish these spatial differences by referring to the inner - halo regions ( ihrs ) and outer - halo regions ( ohrs ) ; the relative dominance of the different stellar populations found within these regions naturally differs from one simulated galaxy to the next , depending on its formation history .
the ihr and ohr are defined by the spatial distributions of the ihps and ohps .
the crossing radius between the density distributions of these two populations is @xmath7 kpc ( tissera et al .
, in preparation ) . to understand their history of assembly , we classify the stars in the ihps and ohps into three sub - populations according to their sites of formation : \(1 ) _ debris _ stars are formed in separate sub - galactic systems located outside the virial radius of the main progenitor galaxies , and its main progenitor as the most massive sub - galactic system from where it formed , as a function of redshift .
] at the time they were born , and were later accreted onto the main galaxy ( i.e. stars from dry mergers as well as the existing populations from wet mergers ) , \(2 ) _ disc - heated _ stars are formed in the disc structure of the main progenitor galaxies , then heated up by some violent event , and \(3 ) _ endo - debris _ stars are those formed in gas - rich sub - galactic systems located within the virial radius of the main progenitor galaxies at their time of formation ( i.e. wet mergers ) . in paper
i , disc - heated and endo - debris stars were classified as the component . in this paper , we distinguish them as separate sub - populations because , at least in our simulations , they exhibit different properties which might help link observations with galaxy formation models .
we note that endo - debris stars could be taken either as part of the population , because they formed within the virial radius of the main progenitor galaxy , or as part of the debris population , because they were born from gas associated with a sub - galactic system .
although we analyse the three sub - populations separately , when required and to be consistent with paper i , we adopt the former interpretation .
note that we have cleaned our haloes from bound substructures that could be isolated in density phase .
these substructures will be considered in a separate paper .
this paper is organized as follows . in section 2 , we describe the simulations and the main aspects of our chemical - enrichment model .
this section also explains the criteria adopted to separate different stellar components . in section 3 ,
we study the chemical and kinematic properties of the different stellar sub - populations contributing to the ihps , in section 4 , we discuss the nature of the ohps .
we summarize our findings in section 5 .
we analyse a suite of six high - resolution simulations of mw galaxy - sized systems of the original sample from the aquarius project , which include the physics of baryons as described by @xcite .
we first selected six out of the eight cases discussed by @xcite , because two haloes experienced important mergers in the recent past , unlike in the mw .
the six selected haloes ( a , b , c , d , g and h ) have surviving discs of different bulge - to - total mass ratios ( see * ? ? ? * and paper i ) .
the target haloes were selected from a cosmological volume consistent with a @xmath5-cdm cosmogony , with parameters @xmath8 and @xmath9 , and @xmath10 ( see * ? ? ?
* ) for further details on the generation of the initial conditions ) .
the simulations were run with a version of gadget-3 @xcite , which includes a multiphase medium , sn energy feedback , and chemical evolution , as described in @xcite .
maximum gravitational softenings in the range @xmath11 kpc @xmath12 were adopted .
virialized structures are identified by requiring the mean overdensity to reach @xmath13 times the cosmological critical density .
the simulated haloes have @xmath14 million total particles within the virial radius , and a virial mass in the range @xmath15 at @xmath6 .
dark matter particles carry masses on the order of @xmath16 , while , initially , gas particles possess a total mass of @xmath17 .
their characteristic parameters are summarized in table 1 and table 2 of ( * ? ? ?
* see also scannapieco et al .
2009 ) .
the version of gadget-3 used to run these simulations includes a multiphase model for the gas component , metal - dependent cooling , star formation , and sn feedback , as described in @xcite and @xcite .
these multiphase and sn - feedback models have been used to successfully reproduce the star - formation activity of galaxies during quiescent and starburst phases , and are able to drive violent mass - loaded galactic winds with a strength reflecting the depth of the potential well @xcite .
this physically - motivated thermal sn - feedback scheme does not include any ad hoc mass - scale - dependent parameters , and has no requirement to switch off cooling or provide a dynamical `` kick '' to other particles . as a consequence ,
it is particularly well - suited for the study of galaxy formation in a cosmological context .
our chemical evolution model includes the enrichment by type ii and type ia supernovae ( snii and snia , respectively ) .
a salpeter initial mass function is assumed , with lower and upper mass cut - offs of 0.1 @xmath18 and 40 @xmath18 , respectively .
we follow 12 different chemical isotopes : h , @xmath19he , @xmath20c , @xmath21o , @xmath22 mg , @xmath23si , @xmath24fe , @xmath25n , @xmath26ne , @xmath27s , @xmath28ca , and @xmath29zn .
initially , baryons are in the form of gas with primordial abundance , x@xmath30 and y@xmath31 .
snii are considered to originate from stars more massive than 8 @xmath32 .
their nucleosynthesis products are adopted from the metal - dependent yields of @xcite .
the lifetimes of snii are estimated according to the metal - and - mass - dependent lifetime - fitting formulae of @xcite . for snia
, we adopt the w7 model of @xcite , which assumes that snia events originate from co white dwarf systems in which mass is transferred from the secondary to the primary star until the chandrasekhar mass is exceeded , and an explosion is triggered . for simplicity , we assume that the lifetime of the progenitor systems are selected at random over the range @xmath33 $ ] gyr . to calculate the number of snia , we adopt an observationally motivated relative ratio of snii to snia rates , as explained by @xcite
. the ejection of chemical elements is grafted onto the sn - feedback model , so that chemical elements are distributed within the cold and hot gas phases surrounding a given star particle .
the fraction of elements going into each phase is regulated by a free parameter , @xmath34 , which also determines the amount of sn energy received by each phase . while the injection of energy follows two different paths , depending on the thermodynamical properties of the gas , chemical elements
are injected simultaneously with the occurrence of the sn event .
all of these simulations were run with @xmath35 . [ cols="<,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] in order to analyse the chemical properties of the stars , and to relate them to the assembly history of the final galaxy , paper i defined different dynamical components by using the binding energy ( @xmath36 ) and angular momentum content along the rotational axis , @xmath37 .
discs are defined by those particles with @xmath38 , where @xmath39 is the maximum @xmath40 over all particles of a given binding energy @xmath36 .
the rest of the stellar particles are taken as part of the spheroidal component .
taking into account their binding energies , stellar particles are classified as belonging to the central spheroid ( ` bulge ' ) or the inner and outer haloes , so that more energetic particles end up in the halo systems while the less energetic particles define the central spheroids .
although the classification of stars belonging to the inner and outer haloes is made on the assumption of a certain binding energy ( as explained in paper i ) , variations of this limit only weakly affect our results , if these variations are sensible . by applying this methodology ,
the dynamical decomposition of the stellar populations of the main progenitor galaxies into discs , central spheroids , ihps and ohps was performed from @xmath41 to @xmath42 for all of the aquarius simulated systems .
we also cleaned the stellar haloes of substructure , as defined by the subfind algorithm , so that the analysed stellar haloes correspond to the diffuse components in density space .
as shown in paper i , the different dynamical components receive contributions from stars formed both _ in situ _ and in sub - galactic systems that were later accreted into the potential well of the progenitor systems .
as defined in the introduction , the component is split into two categories : the disc - heated and endo - debris sub - populations .
the latter are formed in gas - rich sub - systems which could survive farther into the dark matter haloes and continued their star formation activity .
new stars formed in this environment would be more chemically enriched by snia than older ones , particularly if these sub - galactic systems had bursty star formation histories .
the fractions of each stellar sub - population ( debris , endo - debris and disc - heated ) associated with each dynamical component ( ihp or ohp ) is different , and varies from system - to - system , depending on their history of assembly .
paper i also demonstrated that the ohps are mainly formed from debris stars , along with a small fraction of stars . in order to understand the origin of these stars
, we followed each of them back in time , and identified the first sub - galactic system that hosted them .
we found that most of the stars that were originally classified as stars were in fact formed _ in between _
available snapshots of the simulations , which complicated their detection . from seeking the original gas particles in which they formed ,
it is clear that most of them were part of small sub - galactic systems . a negligible fraction of the stars ( and their progenitor gas particles )
could not be related to any system , but all of them are old , with formation redshift @xmath43 . at such high redshifts , our simulations do not possess sufficient resolution to reliably follow the evolution of the smallest clumps , so we include them within the endo - debris populations .
thus , we will hereafter consider that the ohps are built up mainly by debris stars from accreted sub - galactic systems , with a small fraction corresponding to endo - debris stars .
a similar analysis was carried out for the stars in the ihps .
we found that these stars included endo - debris sub - populations as well as disc - heated stars .
the latter are stars formed in the disc components , but that were heated sufficiently to become part of inner haloes by @xmath44 , contributing from @xmath45 to @xmath46 of the ihps .
disc - heated stars are defined by checking if they were formed from cold gas which was part of the disc component of the main progenitor galaxy at the time they were born . to do that we traced each star particle back in time , and examined to which dynamical component it belonged at that time .
this was possible because we defined disc and halo components at every available snapshot of the simulations , as discussed above .
these fractions are in agreement with the recently reported @xmath47 of disc - heated stellar fraction estimated for the mw halo by @xcite .
endo - debris stars contribute @xmath48 of the ihps , on average , and include a small fraction of stars which could not be identified as belonging to any sub - galactic system ( neither could their gaseous progenitors ) .
we also checked for possible contamination of the ihps from very extended thick - disc populations , by relaxing the condition for particles to be classified as part of the disc components at @xmath6 , assuming @xmath49 .
no significant differences were found . in the following sections we analyse the chemical and dynamical properties of the ihps and ohps , as populated by contributions from the three different sub - populations : debris stars , disc - heated stars , and endo - debris stars .
table [ tab1 ] summarizes the percentage of stars in each sub - populations and their median abundances .
we emphasize that , at any given time ( owing to their extended orbits ) , stars that are members of the ihps and ohps can occupy the same spatial regions ( both the ihr and the ohr ) , with different relative fractions at different distances from the centre .
we perform an analysis and further discuss this important point , and its consequences , in tissera et al .
( in preparation ) .
as discussed above , our simulated ihps are complex structures , with contributions from disc - heated stars , as well as from the debris and endo - debris stellar sub - populations .
the debris sub - population is the primary contributor , with an average of @xmath50 across the six simulated haloes , while the endo - debris and disc - heated sub - populations each contribute @xmath51 to @xmath46 , on average . to better visualize the global differences between the ihps in our simulated galaxies , fig .
[ toomregeneral ] ( left panel ) shows the toomre diagram defined by the median velocities , with each of the contributing sub - populations marked .
following @xcite , the velocity coordinates @xmath52 , and @xmath53 are estimated in the rest frame of fiducial solar neighbourhoods defined for the stellar disc component of the simulated galaxies .
the rest frames are defined by estimating the mean velocity components for stars in the stellar discs at @xmath54 kpc .
velocities are normalized to the median rotation velocity of the disc , v@xmath55 , in the fiducial solar neighbourhoods , in order to consider the differences in mass between the simulated galaxies .
hence , the limit between retrograde and prograde motions is set at @xmath56 . as can be seen from inspection of this figure , the debris and endo - debris sub - populations exhibit , on average , little evidence for strong rotation .
typically , disc - heated stars exhibit different kinematics ; most are more rotationally supported than the stars in the other two sub - populations .
however , there exist an important subset of disc - heated stars which can not be distinguished from debris or endo - debris stars .
in fact , there is a clear age dependence on the behaviour of the disc - heated stars
old disc - heated stars ( @xmath57 gyr ; open circles ) do not show significant rotational velocities , while younger stars ( @xmath58 gyr ; filled circles ) exhibit clear signals of rotation .
note that , in these simulations , young disc - heated stars are nevertheless older than @xmath59 gyr . are dominated by young stars as shown by @xcite .
however , discs contributed with heated stars to the ihps mainly at @xmath60 . for this reason ,
the disc - heated stars found in the ihps are older than @xmath61 gyrs . ] in the case of aq - a-5 , a fraction of old disc - heated stars actually exhibit retrograde rotation , illustrating the complexity and diversity of halo assembly in a hierarchical clustering scenario . due to the time delay between the corresponding evolution of the progenitor stars of snii and snia ,
the @xmath2-element - to - iron ratios provide information about the star - formation history of the systems .
[ toomregeneral ] ( middle panel ) shows the median [ o / fe ] vs. [ fe / h ] for the ihps , segregated into the different stellar sub - populations . on average ,
debris stars are @xmath2-enriched relative to iron .
however , we also find a number of stars with low [ @xmath2/fe ] ratios , varying from @xmath62 to @xmath63 , evidencing contributions from snia .
the fraction of such stars correlates with the mass of the accreted sub - galactic systems , so that the larger fractions are found in ihps that formed from more massive systems .
the endo - debris and disc - heated sub - populations have a larger fraction of low ( or even sub - solar ) [ @xmath2/fe ] ratios .
the stellar sub - populations also occupy different locations in the [ o / fe ] vs [ fe / h ] plane .
debris stars are metal poor , and have the smallest contribution from low-[@xmath2/fe ] stars .
endo - debris stars have lower mean [ @xmath2/fe ] ratios .
the disc - heated stars tend to exhibit , on average , both higher metallicities and lower [ @xmath2/fe ] ratios compared to the other sub - populations . as expected , this is particularly true for the younger disc - heated stars .
we found that , on average , @xmath64 of the @xmath2-enriched stars in ihps belong to the debris sub - components , while disc - heated stars and endo - debris stars contribute , individually , on the order of @xmath65 of the total mass fraction of @xmath2-enriched stars , on average .
observationally , there is evidence for the existence of halo stars with low @xmath2-enrichment @xcite , although their frequency is not yet clearly determined .
the endo - debris sub - populations in the ihps could be associated with the low-@xmath2 and low [ feh ] stars classified as the accreted stars by @xcite . in the middle panel of fig .
[ toomregeneral ] , we show the observations of @xcite superposed with the median values estimated for the three sub - populations in our simulated ihps .
observations of the [ @xmath2/fe ] abundance ratios for large , unbiased samples of mw halo stars will provide very important constraints on galaxy formation models . according to our present models , such measurements would be a probe of the complex mixture of stellar sub - populations in the inner haloes of large spiral galaxies such as the mw and m31 .
testable predictions of various formation and evolution scenarios could then be made from which models can be revised and improved . with respect to the mean rotational velocities , the @xmath2-enriched stars followed the same mean trends of their corresponding sub - populations
those belonging to the debris or endo - debris sub - populations exhibit little or no net rotation with respect to the galactic centre in the range @xmath66 $ ] km s@xmath67 , while those classified as disc - heated stars tend to exhibit clear prograde rotation ( @xmath68 km s@xmath67 ) , particularly the younger stars .
the @xmath2-rich stars tend to be @xmath69 gyr older than @xmath2-poor stars
. a portion of the latter coming from the disc components can be up to 1 gyr younger .
+ fig .
[ ages ] shows the ages of the ihps , as a function of binding energy , for the three different stellar sub - components .
the debris stars are the oldest , and they span the entire range of total energies .
these stars are among the first ones to be formed ; as a consequence , they are primarily low metallicity and @xmath2-enriched relative to iron ( as can be appreciated from inspection of fig .
[ toomregeneral ] ) .
the endo - debris stars tend to be slightly younger , and relatively more gravitationally bound .
they formed in gas - rich satellites accreted at @xmath70 ; @xmath71 of them were formed in sub - galactic systems with dynamical masses smaller than @xmath72 m@xmath73 . on average ,
these stars have a larger fraction of low @xmath2-enrichment when compared to the debris sub - populations . as mentioned before , there are two distinctive regions occupied by disc - heated stars
the most tightly bound stars tend to be older and @xmath2-enriched , while the younger disc - heated stars extend to higher energies and are more @xmath2-poor .
our ihps comprise primarily very old stars , with more than @xmath71 older than 10 gyrs ( a lookback time comparable to redshift @xmath74 ) , while , on average , only @xmath75 of them are younger than @xmath76 gyrs ( @xmath77 ) .
most of the latter stars are disc heated .
as mentioned above , the oldest stars tend to dominate at higher energies , thus the proportion of old and young stars might be expected to vary with galactocentric distance .
observations of stellar populations in galactic haloes at higher redshifts should help to constrain their assembly @xcite .
the different chemical abundances and binding energy distributions of the stellar sub - populations can imprint differences onto their metallicity distribution profiles , which , in turn , provides information on the relative contributions of each of the sub - populations .
previous works have examined the metallicity profiles of their simulated stellar spheroids as a whole ( i.e. , central spheroids , inner and outer stellar haloes ; font et al .
2011 ) , or looked at the difference between the inner and the outer stellar haloes @xcite , and reported that these simulations , within the concordance @xmath5cdm scenario , are able to reproduce metallicity trends similar to those reported by observers for the mw and m31 ( see introduction for a discussion on this point ) .
[ gradhaloes ] shows the [ fe / h ] profiles of the disc - heated , endo - debris , and debris stars , as well as the total profiles .
the abundance profiles represent the median [ fe / h ] estimated in equal - size radial intervals ( displayed medians have been estimated with at least 100 stellar particles ) .
the total metallicity profile is clearly a combination of the contributions from each of the sub - populations , the relative importance of which varies from galaxy - to - galaxy .
the resulting metallicity profiles depend on both the level of enrichment of the different stellar sub - populations , and their final spatial density distributions .
the latter has been included in the corresponding insets of fig .
[ gradhaloes ] . hence
, although each of these sub - populations may exhibit weak gradients individually , when they are combined , the outcome could be a steeper profile .
in fact , in some cases , there exists a sharp transition between the region dominated by the sub - populations ( i.e. , disc - heated plus endo - debris stars ) and that dominated by the debris sub - population . in such cases , fitting a simple linear relation might provide spurious results . consequently , we prefer to understand the characteristics of the metallicity profiles by analysing the spatial and binding energy distribution of each stellar sub - population .
this allows us to assess the role played by each sub - population in shaping the final abundance profiles of the ihps .
the eventual aim is to make use of such profiles ( or quantities derived from them ) as tools for describing the assembly histories of individual galaxies . from inspection of fig .
[ gradhaloes ] , there are three ihrs that exhibit steeper variations of [ fe / h ] with distance than the rest . in these cases ,
the sub - components ( disc - heated stars and/or endo - debris stars ) are more centrally concentrated with respect to the debris sub - populations . in the cases with flat [ fe / h ] profiles , the three stellar sub - populations are better mixed .
this can be seen from the inset of fig .
[ gradhaloes ] , where we display the spatial distributions of the three different stellar sub - populations in the six analysed haloes .
it is clear that aq - a-5 , aq - c-5 , and aq - g-5 have their stars more concentrated toward lower energies ( see fig .
[ ages ] ) and the central regions with respect to the debris stars . since the disc - heated stars and , in some cases , the endo - debris stars , have higher metallicities in the central regions than debris stars , the abundance profiles are steeper .
it can be also seen that the debris stars dominate at larger distances . according to our simulations ,
a steep [ fe / h ] profile is the result of poorly mixed sub - populations , with significant contributions from either disc - heated stars , endo - debris stars , or both , in the central regions .
the ohps in our simulated halo systems comprise stars older than @xmath78 gyr , which are very metal poor and @xmath2-enriched relative to iron .
they formed from the accretion of satellite systems , so that most of the stars belong to the debris sub - population .
there is a small fraction of stars that originated from an endo - debris sub - population ( which were in part classified as in paper i ; see see section 2.2 ) . regarding the kinematics
, we find that most of the ohps exhibit low net global rotation with respect to their corresponding galactic centres , with mean values in the range @xmath79 km s@xmath67 .
one of the six ohps ( aq - g-5 ) has a mean @xmath80 km s@xmath67 , and another ( aq - c-5 ) exhibits a retrograde net rotation @xmath81 km s@xmath67 . at least in our simulations ,
the stars are very well - mixed dynamically , so no significant differences in the toomre diagram could be found between the debris stars and the endo - debris stars , or between the low- and high-[@xmath2/fe ] sub - populations ( left panel of fig . [ toomrehaloex ] ) . the middle panel of fig .
[ toomrehaloex ] shows the [ o / fe ] versus [ fe / h ] plane for the debris and endo - debris sub - populations .
it is very clear that the endo - debris sub - populations exhibit a larger contribution from low-[@xmath2/fe ] stars , on average .
however , there is no trend with rotational velocity , as can be seen from the right panel of fig .
[ toomrehaloex ] . the existence ( or absence ) of such stars
might be tested with observations ; their frequency could provide strong constraints on galaxy - formation models and on how the star - formation activity proceeded in small systems at high redshift .
note that most of the stars in the ohps are located outside the solar neighbourhood ; less than @xmath82 can be found within this region ( tissera et al . in preparation ) , thus caution should be taken when compared to available observational surveys .
previous works have reported the existence of a ` gradient ' in the metallicity of stars in the simulated halo systems by considering the difference between the so - called ihrs and ohrs . in paper
i , we measured the difference between the median [ fe / h ] of the ihrs and ohrs , finding that simulations were capable of reproducing values similar to those estimated in the the mw @xcite . however , if the analysis is restricted only to the ohrs , one might expect little or no metallicity gradients , at least if the stars from the accreted sub - galactic systems were well - mixed . in paper
i , we found that the median [ fe / h ] ( [ o / fe ] ) of our ohps are correlated ( anti - correlated ) with the fractions of stars coming from massive satellites , indicating that some memory of the history of formation may still remain in this component .
below we check on this possibility , assuming that their formation histories are consistent with those predicted by a hierarchical clustering scenario .
[ gradhaloex ] shows the median [ fe / h ] and [ o / fe ] profiles for the ohps in our six simulated halo systems . for [ fe / h ]
, there is considerable variation in both the level of enrichment and the predicted slopes with galactocentric distance .
we also note that , although this component has been defined to be smooth , the abundance distributions might still reveal their complex assembly histories as shown by @xmath83 -@xmath36 in paper i ( see also * ? ? ?
this is particularly clear from the [ @xmath2/fe ] profiles plotted in the lower panel of fig .
[ gradhaloex ] , where the contributions from the sub - populations with different levels of @xmath2-enrichment as a function of galactocentric radius can be seen . for the [ fe / h ] profiles ,
a linear regression was fit between @xmath7 kpc and @xmath84 kpc , in order to avoid the central and very outer parts of the distributions , which might be affected by the ihps and the boundaries of the systems , respectively .
[ corre ] shows the slope and zero points , as a function of the fraction of stars formed in more massive sub - galactic systems ( m @xmath85 m@xmath73 ) .
as expected from the results already discussed in paper i , the zero point shows a trend indicating higher metallicities in ohps formed from larger contributions of debris stars acquired from relative more - massive systems .
although the slopes vary from mild to flat , there is a trend with @xmath86 such that steeper negative slopes are found in diffuse ohps formed with significant contributions from relatively higher - mass systems .
massive sub - galactic systems would tend to survive farther into the potential wells , carrying relatively more metal - rich stars and gas available for star formation toward the central regions .
the ohps received a significant contributions of debris from relatively lower - mass satellites will tend to be better - mixed , and exhibit more homogeneous chemical distributions . in order to better understand the differences in the abundance gradients , we also analysed the ages , spatial distributions , and binding energy distributions of the simulated ohps .
we found that the oldest stars are less metal - enriched , as expected , and they span a larger range of total energy and radial distances .
relatively higher - metallicity stars tend to be more gravitationally bound , and are more centrally concentrated .
the outer portions of the ohps are dominated by high - energy stars , which are also the oldest and the lowest metallicity . to illustrate this , fig .
[ binding ] shows the distribution of binding energies for stars in the simulated ohps , segregated according to [ fe / h ] .
the two haloes with the flattest metallicity profiles are those that have similar total energy distributions for stars of different [ fe / h ] abundances . those with the steeper slopes possess relatively higher - metallicity stars predominantly at low total energies . as discussed previously , relatively more - massive satellites contribute higher - metallicity stars .
thus , the existence of a gradient in the ohps is a consequence of the properties of the accreted satellites , and how they were disrupted within the potential well of the systems .
this might explain why analytical models grafted onto the purely dynamical versions of the aquarius systems obtained flatter slopes @xcite .
a comprehensive discussion on the metallicity gradients of both ihps and ohps is considered in tissera et al .
( in preparation ) , together with comparisions with recent observations . accordingly ,
our results indicate that the observed metallicity gradient in the diffuse ohr of a galaxy such as the mw can provide clues about the mass distributions of the satellites that contributed to its formation .
we have analysed a suite of six mw mass - sized systems , simulated within @xmath5cdm using a version of p - gadget3 that includes a prescription for self - consistent sn feedback .
this scheme allows us to follow the chemo - dynamical evolution of baryons as they are assembled into galaxies .
our simulated diffuse stellar haloes can be described as assembled from three stellar sub - populations : debris , endo - debris and disc - heated stars .
debris stars formed in separate sub - galactic systems located outside the virial radius of the main progenitor galaxies at their time of formation .
these sub - galactic systems were later accreted by the main progenitor galaxies , donating stars primary to the ohps , but as well as to ihps .
the endo - debris sub - populations formed in gas - rich sub - galactic systems within the virial radius of the main progenitor galaxies .
these systems were disrupted farther in the potential well , contributing stars primarily to the ihps , and to a lesser degree , to the ohps .
disc - heated stars can represent a significant percentage of the ihps .
these sub - populations are distinguished from one another as they exhibit , on average , stars of different chemical abundances , ages , kinematics , and binding energies .
the relative fractions of debris stars and endo - debris stars provides clues to the mass distribution ( and relative number ) of the sub - galactic systems that contribute to form the stellar haloes .
these results have clear implications for observational studies that attempt to identify different populations in the mw and other galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
summary of the main results : the simulated ihps possess three clear stellar sub - populations the debris stars , the endo - debris stars , and the disc - heated stars .
debris stars are @xmath2-enhanced relative to iron , low metallicity , and on average , show no clear signals for rotation .
endo - debris stars exhibit lower @xmath2-enhancement than debris ones , on average .
endo - debris stars could be associated with the stars observed and classified by @xcite as accreted stars . in our simulations endo - debris stars exhibit larger velocity dispersions , on average , than the disc - heated stars .
disc - heated stars are less @xmath2-enhanced and more metal - rich compared to the endo - debris stars , with characteristics that depend on their ages .
disc - heated stars with ages greater then @xmath87 gyr are dominated by their dispersion , while those with ages in the range @xmath88 gyr have a larger rotational velocity component , and exhibit solar to sub - solar [ @xmath2/fe ] ratios .
although it is feasible , on average , to distinguish these different components by comparing their chemical abundances and kinematic properties , there exists considerable overlap between these sub - populations . the simulated ihps exhibit metallicity gradients , the strength of which are determined by the relative fractions of the three sub - populations that contributed to their formation . while the endo - debris and disc - heated stars dominate in the central parts of ihrs ( within @xmath89 kpc )
, the debris stars dominate in the outer parts of the ihrs .
hence , the relative contributions of the different sub - populations can conspire to produce steeper metallicity gradients than those exhibit by each sub - populations separately .
the simulated ohps are primarily comprised of debris stars that formed in accreted sub - galactic systems .
the relatively less - massive systems are easily disrupted , while the relatively more - massive ones are able to survive farther into the global potential well of the system before being disrupted . as a consequence , although stars of the ohps are generally of very low metallicity , the relatively more metal - rich ones tend to be more centrally concentrated . some of the ohps show signals of weak prograte rotation ( @xmath90 v@xmath55).the relative distribution of debris stars from these different accreted sub - galactic systems can result in a genuine metallicity gradient within the ohrs , which although weak , can be used to constrain the assembly history of the ohps .
the [ @xmath2/fe ] ratios exhibit a heterogeneous distribution from simulated galaxy - to - galaxy , reflecting the mixture of contributions from accreted galactic systems with different star - formation histories .
a future comparison between the frequency of low-[@xmath2/fe ] stars in the haloes can provide clues on the history of formation and the masses of the accreted systems , as well as on the validity of current chemical - enrichment schemes . in this paper
, we have studied the ihps and ohps as separate sub - populations in relation to their assembly histories in a cosmological context .
the properties of the combined ihps and ohps will be discussed in a forthcoming paper , which considers comparisons of the nature of the observed stellar populations in the mw , m31 , and and other nearby galaxies with predictions from the simulations .
we thank simon d.m .
white for useful comments and suggestions .
this work was partially funded by loreal - unesco - conicet award for women in science , pict max planck 245 ( 2006 ) of the ministry of science and technology ( argentina ) , and the cosmocomp and lacegal networks of fp7 programme of the european community .
tcb acknowledges partial funding from grants phy 02 - 16783 and phy 08 - 22648 : physics frontier center / joint institute for nuclear astrophysics ( jina ) , awarded by the u.s
. national science foundation .
j. , freeman k. c. , 2003 , in e. perez , r. m. gonzalez delgado , & g. tenorio - tagle ed . , star formation through time vol .
297 of astronomical society of the pacific conference series , unravelling the epoch of dissipation .
pp 457+ a. s. , benson a. j. , bower r. g. , frenk c. s. , cooper a. , de lucia g. , helly j. c. , helmi a. , li y .- s .
, mccarthy i. g. , navarro j. f. , springel v. , starkenburg e. , wang j. , white s. d. m. , 2011 , mnras , 417 , 1260 | we investigate the chemical and kinematic properties of the diffuse stellar haloes of six simulated milky way - like galaxies from the aquarius project .
binding energy criteria are adopted to define two dynamically distinct stellar populations : the diffuse inner and outer haloes , which comprise different stellar sub - populations with particular chemical and kinematic characteristics .
our simulated inner- and outer - halo stellar populations have received contributions from debris stars ( formed in sub - galactic systems while they were outside the virial radius of the main progenitor galaxies ) and endo - debris stars ( those formed in gas - rich sub - galactic systems inside the dark matter haloes of the main progenitor galaxy ) .
the inner haloes possess an additional contribution from disc - heated stars , in the range @xmath0 , with a mean of @xmath1 .
disc - heated stars might exhibit signatures of kinematical support , in particular among the youngest ones .
endo - debris plus disc - heated stars define the so - called stellar populations . in both
the inner- and outer - halo stellar populations , we detect contributions from stars with moderate to low [ @xmath2/fe ] ratios , mainly associated with the endo - debris or disc - heated sub - populations .
the observed abundance gradients in the inner - halo regions are influenced by both the level of chemical enrichment and the relative contributions from each stellar sub - population .
steeper abundance gradients in the inner - halo regions are related to contributions from the disc - heated and endo - debris stars , which tend to be found at lower binding energies than debris stars . in the case of the outer - halo regions , although [ fe / h ] gradients are relatively mild , the steeper profiles arise primarily due to contributions from stars formed in more massive satellites , which sink farther into the main halo system , and tend to have higher levels of chemical enrichment and lower energies .
our findings support the existence of ( at least ) two distinct diffuse stellar halo populations , as suggested by a number of recent observations in the milky way and m31 .
our results also indicate that a comparison of the range of predicted kinematics , abundance gradients , and frequency of [ @xmath2/fe]-deficient stars with observations of these quantities in the milky way , m31 , and other large spirals can both provide clues to improve the modelling of baryonic physics , and reveal detailed information about their likely history of formation and evolution .
galaxies : haloes , galaxies : structure , cosmology : theory |
many simulation problems in finance and other applied fields can be written in the form @xmath2 , where @xmath3 is a measurable function on @xmath4 and @xmath5 is a standard normal vector , that is , @xmath6 is jointly normal with @xmath7 and @xmath8 .
it is a trivial observation of surprisingly big consequences that @xmath9 for every orthogonal transform @xmath10 .
while this reformulation does not change the simulation problem from the probabilistic point of view , it does sometimes make a big difference when quasi - monte carlo simulation is applied to generate the realizations of @xmath5 .
examples are supplied by the well - known brownian bridge and pca constructions of brownian paths which will be detailed in the following paragraphs .
assume that one wants to know @xmath11 where @xmath12 is a brownian motion with index set @xmath13 $ ] .
in most applications this can be reasonably approximated by @xmath14 , where @xmath15 is a function on the set of discrete brownian paths .
there are three classical methods for sampling from @xmath16 given a standard normal vector @xmath5 , namely the forward method , the brownian bridge construction and the principal component analysis construction ( pca ) .
all of these constructions may be written in the form @xmath17 , where @xmath18 is an @xmath19 real matrix with @xmath20 for example , the matrix @xmath18 corresponding to the forward method is @xmath21 while pca corresponds to @xmath22 , where @xmath23 is the singular value decomposition of @xmath24 .
a corresponding decomposition for the brownian bridge algorithm is given , for example , by larcher , leobacher & scheicher @xcite .
it has been observed by papageorgiou@xcite that @xmath25 if and only if @xmath26 for some orthogonal matrix @xmath27 , so that every linear construction of @xmath16 corresponds to an orthogonal transform of @xmath4 . in that sense
the forward method corresponds to the identity , pca corresponds to @xmath28 and brownian bridge corresponds to the inverse haar transform , see @xcite .
thus our original simulation problem can be written , as @xmath29 with @xmath30 , and we interpret this as using the forward method .
consequently , the same problem using the brownian bridge takes on the form @xmath31 , where @xmath32 is the matrix of the inverse haar transform , and has the form @xmath33 , with @xmath34 , @xmath35 , @xmath36 as above , when pca is used . as an application one can generalize the classical constructions of discrete brownian paths to discrete lvy paths .
see @xcite .
+ there are some theories as to why an orthogonal transform might have the effect to make the problem more suitable for qmc .
caflisch et al .
@xcite introduce the concept of effective dimension of a function : consider a function @xmath37 with finite variance w.r.t .
normal distribution , that is @xmath38 where @xmath6 is a vector of independent standard normal random variables .
then @xmath39 may be written uniquely as the sum of functions @xmath40 , @xmath41 , where @xmath42 depends on the @xmath43-th coordinate only if @xmath44 and where @xmath45 for all @xmath46 and @xmath47 for @xmath48 , using the so - called anova decomposition of @xmath39 .
furthermore it holds @xmath49 the effective dimension in the truncation sense at level @xmath50 is then the smallest integer @xmath51 such that @xmath52 see @xcite .
typically @xmath53 is chosen as @xmath54 .
therefore , a function with effective dimension @xmath51 is one that , in this sense , almost exclusively depends on the first @xmath51 variables and which therefore is more suitable for qmc .
this is confirmed by empirical evidence .
building on the concept of effective dimension of a function , owen @xcite gives definitions of effective dimensions of function spaces , thus connecting the concepts of effective dimension with that of tractability .
now one can turn this around and try to put as much variance as possible to the first few coordinates , by concatenating @xmath39 with a suitable orthogonal transform .
this is what has been done by imai & tan @xcite and what we will do here , using a different approach .
we shall see in section [ sec : numeric ] that empirical evidence also supports the conjectured efficiency of our method .
however , there is also a disadvantage of that approach : the computation of the orthogonal transform has a cost , which is in general of the order @xmath55 . for large @xmath1
this cost is likely to swallow the potential gains from the transform .
we therefore concentrate on orthogonal transforms which have cost of the order @xmath0 .
examples include discrete sine and cosine transform , walsh and haar transform as well as the orthogonal matrix corresponding to the pca , see @xcite .
+ imai & tan @xcite propose an algorithm to find a good orthogonal transform in the sense that it puts as much variance as possible to the first few dimensions .
they propose to take the first order taylor expansion at some point @xmath56 , i.e.@xmath57 then the contribution of the @xmath43-th component of @xmath5 to @xmath58 is given by @xmath59 .
the columns of the orthogonal transform are chosen by solving optimization problems of the form @xmath60 with @xmath61 .
they suggest to perform this optimization only for the first few columns of the matrix @xmath18 . in this paper
we improve on their algorithm in various directions .
in particular we find a good orthogonal transform that is fast in that it can be computed even in linear time .
+ the remainder of the paper is organized as follows .
section [ sec : householder ] reviews basic properties of householder reflections and shows how they can be used to find fast versions of orthogonal transforms which put most variance on the first @xmath51 variables .
the main part of our article , section [ sec : algo ] , describes algorithms for finding fast orthogonal transforms using again householder reflections .
in contrast to the method of imai & tan @xcite we do not rely on differentiability .
this makes the algorithm useful for barrier - type options .
we further provide some theoretical results which indicate why the method serves to reduce the effective dimension .
section [ sec : numeric ] gives some numerical examples where the methods described earlier are applied to examples from finance .
we will see that the new methods described in section [ sec : algo ] are among the best , both with regard to speed and accuracy .
we provide an appendix where we compute certain expectations depending on the maximum of a brownian path .
this is useful for some of the numerical examples .
we recall the definition and basic properties of householder reflections from golub & van loan @xcite . a matrix of the form @xmath62 where @xmath63 ,
is called a _
householder reflection_. the vector @xmath64 is called the defining _
householder vector_. in the following proposition , @xmath65 denotes the first canonical basis vector in @xmath4 , @xmath66 .
a householder reflection have the following properties : 1 .
if @xmath67 is a vector then @xmath68 is the reflection of @xmath69 in the hyperplane @xmath70 . in particular
, @xmath27 is orthogonal and symmetric , i.e. @xmath71 .
2 . given any vector @xmath72 we can find @xmath63 such that for the corresponding householder reflection @xmath27 we have @xmath73 .
the computation of the householder vector uses @xmath74 floating point operations .
the computation of the matrix - vector multiplication @xmath68 uses at most @xmath75 floating point operations .
see chapter 5.1 of golub & van loan @xcite .
our main application of householder reflections is the following : suppose we know that for a given integration problem @xmath2 some orthogonal transform @xmath76 reduces the effective dimension in the truncation sense to @xmath51 , that is , almost all of the variance of @xmath77 is captured by @xmath78 , @xmath79 .
let @xmath80 , that is , @xmath81 is the @xmath82-th column of @xmath76 .
we can find householder reflections @xmath83 such that @xmath84 , @xmath85 as follows : * let @xmath86 be a householder reflection that maps @xmath65 to @xmath87 .
@xmath86 also maps @xmath87 to @xmath65 . since the vectors @xmath81 are orthogonal we have @xmath88 .
* therefore there exists a householder reflection @xmath89 operating on the last @xmath90 coordinates which maps @xmath91 to @xmath92 .
thus @xmath93 , @xmath94 .
* suppose householder reflections @xmath95 have been constructed such that @xmath96 , @xmath97 .
* then there exists a householder reflection @xmath98 operating on the last @xmath99 coordinates which maps @xmath100 to @xmath101 .
then @xmath102 , @xmath103 .
write @xmath104 . by construction
the first @xmath51 columns of @xmath27 coincide with those of @xmath76 . since
, by assumption , @xmath78 capture almost all of the variance of @xmath77 , the same is true for @xmath105 .
but for small @xmath51 the computational cost for computing @xmath106 is of the order @xmath107 , as compared to general matrix - vector multiplication which occurs a cost of order @xmath55 .
+ imai & tan @xcite and wang & sloan @xcite give examples for which they find good orthogonal transforms @xmath76 that reduce the effective dimension .
however they do not specify how those transforms are applied .
we propose to approximate them using the above method
. however , the main topic of this paper is to present transforms that use only one householder reflection .
this will by detailed in the next section .
let @xmath108 be a measurable function with @xmath109 for a standard normal vector @xmath5 .
we want to approximate @xmath3 by a linear function : @xmath110 where @xmath72 and @xmath111 .
this can be done in different ways .
for example , imai & tan @xcite take the first order taylor expansion of @xmath3 .
in contrast , we take a `` linear regression '' approach , i.e. we minimize @xmath112 first order conditions give @xmath113 therefore , minimizes the variance of the difference between @xmath3 and the linear approximation .
so @xmath114 that is , @xmath115 measures the proportion of variance captured by the linear approximation .
now there exists a unique householder reflection @xmath27 that maps @xmath65 to @xmath116 . with this transform
we have @xmath117 and therefore @xmath118 therefore the linear part of the integration problem depends on the parameter @xmath119 alone .
now , if the linear part constitutes a large part of the integration problem then we have succeeded in putting a large fraction of the variance into the first coordinate by composing @xmath3 with @xmath27 .
let @xmath120 be independent standard normal variables .
let @xmath3 be a function @xmath108 . 1 .
@xmath121 for @xmath122 ; 2 .
if @xmath123 define @xmath124 and go to 4 . ; 3 .
else let @xmath27 be a householder reflection that maps @xmath65 to @xmath116 ; 4 .
compute @xmath125 using qmc .
a drawback of the algorithm is that in general the computation of the expectations in step 1 is no easier than the original problem . in some cases
the expectation can be computed explicitly , though usually in that case also the original problem has an explicit solution .
[ ex : sum_prod ] @xmath126 .
it is easily verified that , with @xmath127 denoting the standard normal density , @xmath128 , @xmath129 therefore it holds that @xmath130 let us find out how much of the variance of @xmath131 is captured by @xmath132 : we write @xmath133 .
then @xmath134 where @xmath135 . on the other hand , it is easy to see that @xmath136 and @xmath137
. therefore we get for the variance of @xmath131 @xmath138 let us try some special values that are related to asian options : @xmath139 with @xmath140 , @xmath141 . for this choice
we get @xmath142 , and @xmath143 . for large @xmath1 the sums in equations and
can be approximated by corresponding integrals such that @xmath144 table [ tbl : var - part ] shows the fraction @xmath145 for a few values of @xmath146 , @xmath147 and @xmath148 .
c|cccccccc @xmath149&0.01&0.02&0.03&0.04 + 0.1&0.0025 & 0.0051 & 0.0076 & 0.0101 + 0.2&0.0026 & 0.0051 & 0.0077 & 0.0103 + 0.3&0.0026 & 0.0052 & 0.0078 & 0.0104 + it can be concluded that in this example almost all of the variance of @xmath131 is captured by @xmath119 .
in general we can not expect that @xmath150 can be computed explicitly .
of course it is an option to compute @xmath150 using ( quasi-)monte carlo , though it is unlikely that this will lead to small overall computing times .
but quite frequently , especially in financial applications , a problem can be written in the form , @xmath151 , where @xmath152 can be computed and @xmath153 is some relatively simple function @xmath154 .
[ alg : main ] let @xmath120 be independent standard normal variables .
let @xmath3 be a function @xmath108 , which is of the form @xmath155 where @xmath156 and @xmath157 . 1 .
@xmath158 for @xmath122 ; 2 . if @xmath123 define @xmath124 and go to 4 . ; 3 .
else let @xmath27 be a householder reflection that maps @xmath65 to @xmath116 ; 4 .
compute @xmath125 using qmc . without additional assumptions on the functions @xmath153 and @xmath39 there is no guarantee that @xmath27 gives better convergence .
nevertheless there are practical examples where this algorithm gives excellent results .
[ ex : asianoption ] consider an arithmetic average value option written on some underlying @xmath34 , @xmath159 and @xmath160 here we have @xmath151 , where @xmath161 and @xmath153 is like in example [ ex : sum_prod ] with @xmath162 , @xmath163 , @xmath164 , @xmath165 .
+ write @xmath166 , @xmath167 .
then @xmath168 are uncorrelated , @xmath169 further , @xmath170 , such that @xmath171 as well , and therefore @xmath172 .
[ th : decomp ] let @xmath173 be like in algorithm [ alg : main ] .
write again @xmath166 , @xmath167
. then @xmath174 .
we write @xmath175 and @xmath176 , so that we have to show @xmath177 . to that end
it is sufficient to prove that @xmath178 and @xmath179 are uncorrelated : @xmath180 since @xmath181 , we have @xmath182 .
we consider a special case that will rarely occur in practice but which gives a flavor of the best result possible .
assume that @xmath39 is lipschitz continuous with constant @xmath183 .
suppose further that @xmath184 and @xmath185 are not only uncorrelated , but even independent .
denote by @xmath186 , @xmath187 the cumulative probability distribution functions of @xmath184 and @xmath185 , respectively .
using independence we get @xmath188 noting that @xmath189 we thus get @xmath190 where we also have used the cauchy - schwarz inequality . thus with theorem [ th : decomp ] we get @xmath191 that is , @xmath192 so in this situation , if @xmath119 captures a large fraction of the variance of @xmath193 , then @xmath119 also captures a large fraction of the variance of @xmath131 provided that the lipschitz constant @xmath183 is not too big .
we can also think of a variant of algorithm [ alg : main ] for slightly more complicated functions .
we have been inspired by wang & sloan @xcite , where the authors consider functions of the form @xmath194 and show , that there is an orthogonal transform that makes this function @xmath195-dimensional .
we give a slightly modified version of their argument which guarantees that the orthogonal transform is also fast to compute for small @xmath195 , that is for @xmath196 .
let @xmath194 for @xmath197 .
we may assume that @xmath198 is not the zero vector .
let @xmath86 be a householder reflection which maps @xmath199 to @xmath200 .
then @xmath201 and therefore @xmath202 next we write @xmath203 .
that is , @xmath204 for some @xmath205 .
assuming that @xmath206 , let @xmath207 be the householder reflection of @xmath208 that maps @xmath199 to @xmath209 and let @xmath210 then @xmath89 is a householder reflection of @xmath4 and @xmath211 for some @xmath212 . proceeding that way
one arrives at @xmath213 for some @xmath214 ( we may have @xmath215 if at some stage all remaining @xmath216 are zero ) . + we propose a similar procedure for an integration problem of the form @xmath217 where @xmath218 can be computed explicitly ( or at least efficiently ) .
[ alg : general ] let @xmath120 be independent standard normal variables .
let @xmath3 be a function @xmath108 , which is of the form @xmath155 where @xmath219 and @xmath220 . 1
. start with @xmath221 and @xmath124 ; 2 .
[ it : goto ] @xmath222 for @xmath223 ; 3 .
@xmath224 for @xmath225 ; 4 .
if @xmath226 define @xmath227 and go to [ it : final ] ; 5 .
else let @xmath228 be a householder reflection that maps @xmath229 to @xmath230 ; 6 .
@xmath231 ; 7 . @xmath232 ; 8 .
while @xmath233 , go back to [ it : goto ] ; 9 .
[ it : final ] compute @xmath125 using qmc
. we will give a numerical example in section [ sec : numeric ] .
in this section we will apply our method to examples from mathematical finance .
the first numerical example we give is the evaluation of an asian call option with discrete arithmetic average in the black - scholes model , which has been discussed previously .
since the payoff function @xmath3 is of the form @xmath234 with @xmath39 and @xmath153 as in example [ ex : asianoption ] , we apply algorithm [ alg : main ] to the integration problem @xmath235 where the vector @xmath236 follows from example [ ex : sum_prod ] , i.e. for every @xmath237 @xmath238 for the quasi - monte carlo simulation we use a sobol sequence of dimension @xmath239 with a random shift and we have @xmath240 as well as @xmath148 .
we compute the standard deviation based on @xmath241 batches for @xmath242 sample paths , where the number of sample paths ranges from @xmath243 to @xmath244 .
note that the standard deviation is different from the rqmc standard deviation defined in lecuyer & munger @xcite . in figure
[ fig : asian ] we compare the regression method with the forward method , the pca construction and the lt method of imai and tan .
we see that pca , lt and the regression method yield similar results , but all of the three outperform the forward method .
note that the regression method can be applied in @xmath245 .
thus we can achieve the efficiency of the pca with the regression method with lower computational costs .
moreover , it is interesting that the lt method and regression method yield nearly the same results
. runs on a @xmath246-scale , width=298,height=207 ] runs on a @xmath246-scale , width=298,height=207 ] the computation time required to price the asian option using quasi - monte carlo integration with @xmath244 paths is given in table [ tbl : comptime ] .
note that pca is implemented using the discrete sine transform as discussed in leobacher @xcite .
the lt method is implemented such that only the first @xmath247 columns are optimized and then the orthogonal transform is completed using householder reflections as we suggested in section [ sec : householder ] .
c|cccc & & & & regression + & 0.08 & 0.64 & 1.94 & 0.15 + furthermore it should be mentioned that the regression method as well as the lt method produce an overhead caused by determining the orthogonal transform .
nevertheless the overhead time is rather small and is negligible for a large sample size .
the computation times of the subsequent numerical examples are similar to the result regarding the asian option .
we consider an asian basket call option with arithmetic average and a basket consisting of @xmath195 assets , an example taken from imai & tan @xcite .
the @xmath43-th asset @xmath248 of the basket ( @xmath249 ) is given by @xmath250 where @xmath251 is the current price of the @xmath43-th asset , @xmath146 is the risk - free interest rate , @xmath252 is the volatility of the @xmath43-th asset and @xmath253 is an @xmath195-dimensional brownian motion .
the correlation between @xmath254 and @xmath255 is denoted by @xmath256 .
the payoff function of the asian basket option is given by @xmath257 where @xmath258 and where @xmath259 is an @xmath260matrix with @xmath261 and @xmath262 is an @xmath263matrix with @xmath264 for all @xmath43 and @xmath265 for @xmath266 .
note that the discussion of the previous sections also holds for a discrete brownian path with covariance matrix @xmath267 .
since the problem is of the form @xmath151 , algorithm [ alg : main ] can be applied .
since the function @xmath153 is of the form considered in example [ ex : sum_prod ] , we can compute the corresponding vector @xmath236 analytically . furthermore , notice that the pca construction can be computed in this example efficiently by using the orthogonal transform @xmath268 where @xmath269 and @xmath270 .
the parameters are @xmath271 and @xmath272 for @xmath273 .
moreover , the volatility of the @xmath274 assets is equally spaced from @xmath275 to @xmath276 and we assume that @xmath277 for all @xmath249 . since we simulate every asset at @xmath278 time points , we take a sobol sequence in dimension @xmath279 with a random shift . in figure
[ fig : asian ] we can observe the standard deviation based on @xmath241 batches of the forward method , the pca construction , the lt method and the regression method for @xmath242 sample paths with @xmath242 up to @xmath244 . a digital ( up - and - in ) barrier option is a derivative which pays @xmath280 if the underlying asset breaks through a barrier @xmath281 on the time interval @xmath13 $ ] and pays @xmath282 otherwise .
we intend to price the option in a discrete black - scholes model , where the path of the stock is given by @xmath283 with @xmath284 with current stock price @xmath285 , interest rate @xmath146 , volatility @xmath147 , brownian path @xmath286 where @xmath287 and standard normal vector @xmath288 .
hence , the payoff function @xmath153 of the digital barrier option is @xmath289 which leads us to an integration problem of the form @xmath290 .
we can use algorithm [ alg : main ] for solving this problem and therefore , we have to compute @xmath291 for @xmath292 . in the appendix
we show how to calculate this expectation for a function depending on the maximum of a brownian motion with drift @xmath293 .
we can adjust our problem by @xmath294 with @xmath295 , @xmath296 and @xmath297 .
with @xmath298 we get that the vector @xmath236 in algorithm [ alg : main ] can be approximated by @xmath299 where @xmath34 is given by ( [ eq : summation - matrix ] ) , @xmath300 with @xmath301 , @xmath302 and @xmath303 .
the computation of @xmath304 with @xmath292 can be reduced to a @xmath280-dimensional integration problem using ( [ eq : refl3 ] ) with @xmath305 and @xmath306 and formula ( [ eq : refl1 ] ) with @xmath307 simplifies @xmath308 .
consequently , we end up with @xmath280-dimensional integrals which can be evaluated efficiently with an adaptive quadrature rule . for the numerical test we use a sobol sequence of dimension @xmath309 with a random shift and the parameter set
is chosen as @xmath310 and @xmath148 . the number of sample paths @xmath242 ranges from @xmath243 to @xmath244 and we compute the standard deviation for those @xmath242 based on @xmath241 batches . since it is not clear how to apply the lt method of imai and tan to barrier options , we compare the regression method with the forward method and the pca construction only . in figure
[ fig : barrier ] we can observe that the difference between the forward method and the pca is smaller than in the previous examples .
furthermore , we see that the regression method is slightly behind the pca , but this seems to be the best we can achieve by linear approximation
. runs on a @xmath246-scale , width=298,height=207 ] runs on a @xmath246-scale , width=298,height=207 ] the last example we provide is an asian ( up - and - in ) barrier option by which we mean that the payoff of the option is similar to an asian option as in the first numerical example , but is paid only if the underlying asset breaks through an upper barrier @xmath281 . the corresponding function
is then given by @xmath311 where @xmath312 is as in ( [ eq : discretestock ] ) for @xmath313 .
since the function @xmath3 is of the form @xmath314 with @xmath315 , @xmath316 and @xmath317 , we apply algorithm [ alg : general ] with @xmath318 to the problem .
the computation of the vectors @xmath319 and @xmath320 is already discussed in the examples above , i.e. @xmath319 is related to the digital barrier option and @xmath320 corresponds to the asian option .
the numerical test is based on @xmath241 batches and we again compare the standard deviation of the forward method , the pca construction and the regression method for various numbers of sample paths @xmath242 , ranging from @xmath243 to @xmath244 .
moreover , we use a sobol sequence in dimension @xmath321 with a random shift and the parameters are @xmath322 and @xmath148 .
figure [ fig : barrier ] shows that the regression method yields slightly better results than the pca and that the forward method is behind the other two approaches .
we give the computations needed for examples of barrier type , that is we want to compute @xmath152 where @xmath153 is some function of the maximum of a discrete brownian path with drift @xmath293 , i.e. @xmath323 and where @xmath324 .
we make the approximation @xmath325 where @xmath326 denotes brownian motion with drift @xmath327 , i.e. @xmath328 , @xmath329
. moreover , let @xmath330 and @xmath331 .
at first we compute @xmath332 for given @xmath333 and measurable @xmath3 with @xmath334 .
then we show how the expectation for more general @xmath335 can be computed using the first result .
we start with a simple calculation for a brownian motion @xmath12 with drift 0 and let @xmath336 .
for @xmath337 we get , using the reflection principle for brownian motion , @xmath338 next we make a girsanov - type change of measure such that under the new measure @xmath339 the brownian motion @xmath326 with drift becomes a standard brownian motion .
so with @xmath340 , that is @xmath341 @xmath342 the next step is to consider @xmath332 for @xmath343 .
let @xmath344 denote the standard filtration of @xmath12 .
@xmath345 we have already computed the first term . for the second term
we note that by the markov property of brownian motion , @xmath346 we can use our earlier result ( [ eq : refl1 ] ) with @xmath347 to obtain @xmath348 let us write @xmath349 .
then , using ( [ eq : refl1 ] ) and ( [ eq : refl2 ] ) we obtain @xmath350 note that the expectations can be computed explicitly for suitable @xmath3 . p. lecuyer and d. munger . on figures of merit for randomly - shifted lattice rules . in h.
woniakowski and l. plaskota , editors , _ monte carlo and quasi monte carlo methods 2010 _ , berlin , 2012 .
springer.to appear .
p. lecuyer , j .- s .
parent - chartier , and m. dion .
simulation of a lvy process by pca sampling to reduce the effective dimension . in _ proceedings of the 2008 winter simulation conference _ , pages 436443 .
ieee press , 2008 . | there are a number of situations where , when computing prices of financial derivatives using quasi - monte carlo ( qmc ) , it turns out to be beneficial to apply an orthogonal transform to the standard normal input variables .
sometimes those transforms can be computed in time @xmath0 for problems depending on @xmath1 input variables . among those are classical methods like the brownian bridge construction and principal component analysis ( pca ) construction for brownian paths .
building on preliminary work by imai & tan @xcite as well as wang & sloan @xcite , where the authors try to find optimal orthogonal transform for given problems , we present how those transforms can be approximated by others that are fast to compute .
we further present a new regression - based method for finding a householder reflection which turns out to be very efficient for a wide range of problems .
we apply these methods to several very high - dimensional examples from finance . |
recently chaotic dynamical systems are very popular in science and engineering . besides the original definition of li - yorke chaos in @xcite ,
there have been various definitions for `` chaos '' in the literature , and the most often used one is given by devaney in @xcite .
although there is no universal definition for chaos , the essential feature of chaos is sensitive dependence on initial conditions so that the eventual behavior of the dynamics is unpredictable .
the theory and methods of chaotic dynamical systems have been of fundamental importance not only in mathematical sciences , but also in physical , engineering , * biological * , and even economic sciences . in this paper ,
a chaos would be understood in the sense of li - yorke ( see @xcite ) . in this short communication , we introduce and examine a family of nonlinear discrete dynamical systems that naturally occurs to describe a transmission of a trait from parents to their offspring . here , we shall present some essential analytic and numerical results on dynamics of such models by avoiding some technical parts of the proofs . in upcoming papers , we shall explore the deepest investigation on asymptotic behaviors of the family of nonlinear dynamical systems by providing the proofs in details . as the first example
, we consider a mendelian inheritance of a single gene with two alleles @xmath0 and @xmath1 ( see @xcite ) .
let an element @xmath2 represent a gene pool for a population and it have been expressed a linear combination of the alleles @xmath3 and @xmath1 @xmath4 where , @xmath5 and @xmath6 then , @xmath7 are the percentage of the population which carries the alleles @xmath3 and @xmath8 respectively . the rules of the mendelian inheritance indicate that the next generation @xmath9 represents a gene pool for the population which carries the alleles @xmath3 and @xmath8 respectively , where @xmath10 here , @xmath11 ( resp .
@xmath12 ) is the probability that the child receives the allele @xmath3 ( resp .
@xmath8 ) from parents with the allele @xmath3 ; @xmath13 ( resp .
@xmath14 ) is the probability that the child receives the allele @xmath3 ( resp .
@xmath8 ) from parents with the alleles @xmath3 and @xmath8 respectively ; and @xmath15 ( resp .
@xmath16 ) is the probability that the child receives the allele @xmath3 ( resp .
@xmath8 ) from parents with allele @xmath8 .
it is evident that @xmath17 thus , is a nonlinear dynamical system acting on the one dimensional symplex @xmath18 that describes the distribution of the next generation which carries the alleles @xmath3 and @xmath8 respectively , if the distribution of the current generation are known .
recall that in the mendelian inheritance case , i.e. , @xmath19 and @xmath20 , the dynamical system has the following form @xmath21 we assume that prior to a formation of a new generation each gene has a possibility to mutate , that is , to change into a gene of the other kind . specifically , we assume that for each gene the mutation @xmath22 occurs with probability @xmath23 and @xmath24 occurs with probability @xmath25 we assume that `` _ _ the mutation occurs if only if both parents have the same allele _ _ '' . then , we have that @xmath26 , @xmath27 and the dynamical system has the following form @xmath28 an operator @xmath29 given by is called a quadratic stochastic operator .
we introduce some standard terms in the theory of a discrete dynamical system @xmath30 . a sequence @xmath31 is called a trajectory of @xmath32 starting from an initial point @xmath33 where @xmath34 for any @xmath35 .
recall that a point @xmath36 is called a fixed point of @xmath32 if @xmath37 we denote a set of all fixed points by @xmath38 .
a dynamical system @xmath32 is called regular if a trajectory @xmath31 converges for any initial point @xmath36 .
note that if @xmath32 is regular , then the limiting point is a fixed point of @xmath39 thus , in a regular system , the fixed point of dynamical system @xmath32 describes a long run behavior of the trajectory of @xmath32 starting from any initial point .
the biological treatment of the regularity of dynamical system @xmath32 is rather clear : in a long run time , the distribution of species in the next generation coincide with distribution of species in the current generation , i.e. , stable .
a fixed point set and an omega limiting set of quadratic stochastic operators ( qso ) were deeply studied in @xcite-@xcite , and they play an important role in many applied problems @xcite . in the paper @xcite , it was given a long self - contained exposition of recent achievements and open problems in the theory of quadratic stochastic operators .
a dynamical system @xmath40 is said to be ergodic if the limit @xmath41 exists for any @xmath42 . based on some numerical calculations
, s.ulam has conjectured @xcite that any qso @xmath43 acting on the finite dimensional simplex is ergodic .
however , in @xcite , m.zakharevich showed that , in general , ulam s conjecture is false .
namely , m.zakharevich showed that the following qso @xmath44 is not ergodic @xmath45 in @xcite , zakharevich s result was generalized in the class of volterra qso .
moreover , in @xcite , it was given a necessary and sufficient condition of being non - ergodicity of volterra qso defined on @xmath46 we define the @xmath47-th order cesaro mean by the following formula @xmath48 where @xmath49 and @xmath50 .
it is clear that the first order cesaro mean @xmath51 is nothing but @xmath52 . in this manner ,
zakharevich s result says that the first order cesaro mean @xmath53 of the trajectory of operator diverges for any initial point except fixed points .
surprisingly , in @xcite , it was shown that any order cesaro mean @xmath54 , where @xmath49 , of the trajectory of operator diverges for any initial point except fixed points .
this leads to a conclusion that the operator @xmath55 given by might have unpredictable behavior .
in fact , in @xcite , it was proven that an operator @xmath55 given by exhibits the li - yorke chaos .
it is worth pointing out that some other properties of volterra qso were studied in @xcite .
note that if qso is regular , then it is ergodic .
however , the reverse implication is not always true .
it is known that the dynamical system is either regular or converges to a periodic-2 point ( see @xcite ) .
therefore , in 1d simplex , any qso is ergodic . in other words , a mutation in population system
having * a single gene with two alleles * always exhibits an ergodic behavior ( or almost regular ) .
it is of independent interest to study a mutation in population system having * a single gene with three alleles*. in the next section , we consider an inheritance of a single gene with three alleles @xmath56 and @xmath57 and show that a nonlinear dynamical system corresponding to the mutation exhibits a non - ergodic behavior ( or li - yorke chaos ) .
in this section , we shall derive a mathematical model of a inheritance of a single gene with three alleles . in this case ,
an element @xmath36 represents a population if its expression @xmath58 , as a linear combination of the alleles @xmath56 and @xmath57 , satisfies the following conditions @xmath59 and @xmath60 then @xmath61 are the percentage of the population which carries the alleles @xmath56 and @xmath57 respectively .
we assume that prior to a formation of a new generation each gene has a possibility to mutate , that is , to change into a gene of the other kind .
we assume that the mutation occurs if _ both parents have the same alleles_. specifically , we will consider two types of the simplest mutations : 1 .
assume that mutations @xmath62 , @xmath63 , and @xmath64 occur with probability @xmath65 ; 2 .
assume that mutations @xmath66 , @xmath67 , and @xmath68 occur with probability @xmath69 then the corresponding dynamical systems are defined on the two - dimensional simplex @xmath70 in the first mutation , we have @xmath71 in the second mutation , we have @xmath72 in both cases , if @xmath73 , i.e. , if a mutation does not occur , then a dynamical systems and coincide with zakharevich s operator . as we already mentioned , zakharevich s operator exhibits a li - yorke chaos ( see @xcite ) .
let @xmath74 . in the first case ,
an operator @xmath75 is a permutation of zakharevich s operator .
an omega limiting set of the permutation of zakharevich s operator was studied in @xcite . by means of results @xcite , @xcite , @xcite , and @xcite
, one can show that the operator @xmath75 is non - ergodic as well as a li - yorke chaos .
let @xmath74 .
in the second case , an operator @xmath76 is a permutation of an operator which was studied in @xcite . by applying the same method which was given in @xcite
, we may show that the operator @xmath76 is regular .
it is easy to check that @xmath77 and @xmath78 . in other words , in the first case , a mutation operator is a convex combination of two li - yorke chaotic operators and in the second case , a mutation operator is a convex combination of a li - yorke chaotic and regular operators .
both operators @xmath79 and @xmath80 were not studied in either paper @xcite or @xcite . in the next section
, we are going to present some essential analytic and numerical results on dynamics of @xmath79 and @xmath80 , by avoiding some technical parts of the proofs . in upcoming papers , we shall explore the deepest investigation on asymptotic behaviors of @xmath79 and @xmath80 by providing the proofs in details .
in this section , we shall study dynamics of the operators and . here , we shall present some pictures of attractors of the operators and . we are aiming to present some analytic results on dynamics of @xmath81 : @xmath82 where @xmath83 and @xmath84 .
as we already mentioned , this operator can be written in the following form : @xmath77 for any @xmath84 , where @xmath85 let @xmath86 be a permutation .
the proofs of the following results are straightforward .
[ permutationandv ] let @xmath87 be a quadratic stochastic operator given by , where @xmath88 .
let @xmath89 and @xmath90 be sets of fixed points and omega limiting points of @xmath79 , respectively . then the following statements hold true .
* operators @xmath91 and @xmath79 are commutative , i.e. , @xmath92 ; * if @xmath93 then @xmath94 ; * if @xmath89 is a finite set then @xmath95 ; * one has that @xmath96 , for any @xmath97 . we are aiming to study the fixed point set @xmath89 , where @xmath88 .
it is worth mentioning that @xmath98 and @xmath99 , where @xmath100 are vertexes of the simplex @xmath101 and @xmath102 is a center of the simplex @xmath101 .
recall @xcite that a fixed point @xmath103 is non - degenerate if and only if the following determinant is nonzero at the fixed point @xmath104 : @xmath105 [ uniquefixedpoint ] let @xmath87 be a quadratic stochastic operator given by , where @xmath88 .
let @xmath102 be a center of the simplex @xmath101 .
then the following statements hold true : * all fixed points are non - degenerate ; * one has that @xmath106 .
let @xmath93 be a fixed point .
one can easily check that @xmath107 if @xmath108 then the above expression is positive .
this means that , in the case @xmath108 , all fixed point are non - degenerate .
let us consider the case @xmath109 in this case , the above expression is equal to zero if and only if @xmath110 .
since @xmath111 , we have that @xmath112 without loss of generality , we may assume that @xmath113 ( see proposition [ permutationandv ] ( i ) ) .
let @xmath114 .
since @xmath93 , we have that @xmath115 we then obtain that @xmath116 ^ 2+x_3 ^ 2\\ & \leq&x_1 ^ 2+[(1-\alpha)x_2+\alpha x_3]^2+x_3 ^ 2\\ & < & x_1
^ 2+x_2 ^ 2+x_3 ^ 2=\frac{1 + 2\alpha^2}{2(1-\alpha+\alpha^2)}\end{aligned}\ ] ] this is a contradiction . in the similar manner , one can have a contradiction whenever @xmath117 .
this shows that , in the case @xmath118 , all fixed points are non - degenerate .
we want to show that @xmath106 .
it is clear that @xmath119 .
this means that the operator @xmath79 does not have any fixed point on the boundary @xmath120 of the simplex @xmath101 , i.e. , @xmath121 .
moreover , all fixed point are non - degenerate . due to theorem 8.1.4 in @xcite , @xmath122 should be odd .
on the other hand , due to corollary 8.1.7 in @xcite , one has that @xmath123 .
proposition [ permutationandv ] , ( iii ) yields that @xmath124 .
simple calculations show that @xmath106 .
we can easily check a local behavior of the fixed point @xmath102 .
[ localbehavioroffixedpoint ] let @xmath87 be a quadratic stochastic operator given by , where @xmath88 .
then the following statements hold true : * if @xmath125 then the fixed point @xmath102 is repelling ; * if @xmath126 then the fixed point @xmath102 is non - hyperbolic . we shall separately study two cases @xmath125 and @xmath126 .
let @xmath87 be a quadratic stochastic operator given by , where @xmath125 .
then @xmath127 is an infinite compact subset of @xmath128 for any @xmath129 .
let @xmath125 .
since @xmath79 is continuous and @xmath130 , an omega limiting set @xmath90 is a nonempty compact set and @xmath127 , for any @xmath129 .
we want to show that @xmath90 is infinite for any @xmath131 since @xmath132 is repelling , we have that @xmath133 .
let us pick up any point @xmath134 from the set @xmath90 . since the operator @xmath79 does not have any periodic point
, the trajectory @xmath135 of the point @xmath136 is infinite .
since @xmath79 is continuous , we have that @xmath137 this shows that @xmath90 is infinite for any @xmath131 it is worth mentioning that the sets of omega limiting points @xmath138 and @xmath139 of both operators @xmath55 and @xmath75 are infinite .
however , unlike the operator @xmath79 , we have inclusions @xmath140 and @xmath141 . moreover , both operators @xmath55 and @xmath75 are non - ergodic .
let @xmath87 be a quadratic stochastic operator given by , where @xmath125 .
then the following statements hold true : * the operator @xmath79 is non - ergodic ; * the operator @xmath79 exhibits a li - yorke chaos .
now , we shall study the case @xmath142 the operator @xmath143 takes the following form @xmath144 in this case , the fixed point @xmath102 is non - hyperbolic and the spectrum of the jacobian of the operator @xmath145 at the fixed point @xmath132 is @xmath146 .
let us define the following sets @xmath147 we have the following cycles : * @xmath148 * @xmath149 let @xmath145 be an operator given by .
one can easily check that @xmath150 the proof the proposition follows from the above equality .
let @xmath151 be a quadratic stochastic operator given by .
the following statements hold true : * @xmath152 is a lyapunov function ; * the trajectory always converges to the fixed point @xmath102 .
let @xmath145 be an operator given by .
it follows from that @xmath153 on the other hand , we have that @xmath154 therefore , one has that @xmath155 for any @xmath156 .
this means that @xmath157 is decreasing a long the trajectory of @xmath145 .
consequently , @xmath157 is a lyapunov function .
we know that @xmath158 is a decreasing bounded sequence . therefore , the limit @xmath159 exists .
we want to show that @xmath160 .
suppose that @xmath161 .
it means that @xmath162 . since @xmath163
, we get that @xmath164 on the other hand , since @xmath162 , there exists @xmath165 such that for any @xmath166 one has that @xmath167 this is a contradiction .
it shows that @xmath160 .
therefore , @xmath168 we want to show that @xmath169 . we know that @xmath170 .
it follows from that @xmath171 this means that @xmath172 .
this completes the proof .
we are going to present some pictures of attractors ( an omega limiting set ) of the operator @xmath81 given by . in the cases @xmath73 and @xmath74 , the operators @xmath55 and @xmath75 have similar behaviors .
the trajectories of both operators , @xmath55 and @xmath75 , look as spirals , but one of them is moving by clockwise and another one is moving by anticlockwise . in these cases , we have that @xmath140 and @xmath141 .
we are interested in the dynamics of the mutation operator @xmath79 while @xmath65 approaches to @xmath173 from both left and right sides . in order to see some symmetry
, we shall provide attractors of @xmath79 and @xmath174 at the same time .
if we slightly change @xmath65 from @xmath175 and @xmath176 , then we can see that the omega limiting set splits from the boundary .
moreover , in the picture , we can see * `` three vertexes '' * ( it is roughly speaking ) where the trajectory shall spend almost all its time around those points .
therefore , we are expecting non - ergodicity of the operator @xmath79 if @xmath65 is near by @xmath175 or @xmath176 .
if @xmath65 becomes close to the @xmath173 then we can see some chaotic pictures .
we observe from the pictures that , in the cases @xmath65 and @xmath177 , the attractors are the same but different from the orientations .
one of them is moving by clockwise and another one is by anticlockwise .
if @xmath65 is enough close to @xmath173 then we detect completely different pictures . here
are some pictures for the values of @xmath178 . in this cases ,
attractors are disconnected sets and they consist of 6 connected sets .
it is very surprising that why the number of connected sets are 6 and why not 5 or 7 .
[ figalpha0109 ] : @xmath179 and @xmath180,title="fig:",width=151,height=151 ] : @xmath179 and @xmath180,title="fig:",width=151,height=151 ] [ figalpha04970503 ] : @xmath181 and @xmath182,title="fig:",width=151,height=151 ] : @xmath181 and @xmath182,title="fig:",width=151,height=151 ] [ figalpha04990501 ] : @xmath183 and @xmath184,title="fig:",width=151,height=151 ] : @xmath183 and @xmath184,title="fig:",width=151,height=151 ] [ figalpha0499505005 ] : @xmath185 and @xmath186,title="fig:",width=151,height=151 ] : @xmath185 and @xmath186,title="fig:",width=151,height=151 ] [ figalpha0499905001 ] : @xmath187 and @xmath188,title="fig:",width=151,height=151 ] : @xmath187 and @xmath188,title="fig:",width=151,height=151 ] for the operator , the fluctuation point is @xmath126 . in this case , the influence of the chaotic operators @xmath55 and @xmath75 are the same . therefore , the operator @xmath189 becomes regular and ergodic .
this completes the numerical study of the operator .
we are aiming to present some analytic results on dynamics of @xmath190 : @xmath191 where @xmath192 and @xmath84 .
as we already mentioned , this operator can be written in the following form : @xmath78 for any @xmath84 , where @xmath193 as we already stated , the operator @xmath194 is zakharevich s operator and the operator @xmath76 is a permutation of the operator which was studied in @xcite . by means of methods which were used in @xcite , we can easily prove the following result .
let @xmath195 be a quadratic stochastic operator given by with @xmath74 .
then the following statements hold true : * the operator @xmath76 has a unique fixed point @xmath196 which is attracting ; * the vertexes of the simplex @xmath100 are 3-periodic points ; * @xmath197 is a lyupanov function ; * the operator @xmath76 is regular in @xmath128 . by means of the same methods and techniques which are used for the operator @xmath79 , we can prove the following results let @xmath198 be a quadratic stochastic operator given by .
then it has a unique fixed point @xmath196 , i.e. , @xmath199 .
moreover , one has that : * if @xmath200 then the fixed point is repelling ; * if @xmath201 then the fixed point is attracting ; * if @xmath202 then the fixed point is non - hyperbolic . let @xmath198 be a quadratic stochastic operator given by
. then the following statements hold true : * if @xmath200 then @xmath127 is an infinite compact set for any @xmath129 ; * if @xmath203 then @xmath204 for any @xmath97 .
let @xmath198 be a quadratic stochastic operator given by , where @xmath200 .
then the following statements hold true : * the operator @xmath80 is non - ergodic ; * the operator @xmath80 exhibits a li - yorke chaos .
we are going to present some pictures of attractors ( an omega limiting set ) of the operator @xmath190 given by . in this cases @xmath73 and @xmath74 , the operator @xmath205 is a chaotic operator and the operator @xmath76 is regular . since @xmath78 , the mutation operator @xmath80 gives a transition from the chaotic behavior to the regular behavior .
consequently , we are aiming to find * fluctuation points * in which _ we can see the transition from the chaotic behavior to the regular behavior . _ [ wfigalpha00101 ] : @xmath206 and @xmath207,title="fig:",width=151,height=151 ] : @xmath206 and @xmath207,title="fig:",width=151,height=151 ] if @xmath65 is very close to @xmath175 then attractors of the operator @xmath80 are splitted from the boundary of the simplex . however , the influence of the operator @xmath205 is still very high .
therefore , we are expecting that the operator @xmath80 is non - ergodic and chaotic , whenever @xmath65 is very close to @xmath175 .
if @xmath65 becomes to close to @xmath208 then we can see a different picture . in attractors
, we are able to detect some * `` growing hairs '' * ( it is a just terminology ) . moreover ,
if we shall continue to increasing @xmath65 then these `` hairs '' will start to rise and they eventually become straight . therefore , * @xmath209 is the first fluctuation point .
* it is very interesting that the number of `` hairs '' is equal to 12 .
it is again a question : why do we have 12 numbers of `` hears '' and why not 11 or 13 .
[ wfigalpha01333 ] : @xmath209 and @xmath210,title="fig:",width=151,height=151 ] : @xmath209 and @xmath210,title="fig:",width=151,height=151 ] [ wfigalpha01390144015 ] : @xmath211 , @xmath212 , and @xmath213,title="fig:",width=151,height=151 ] : @xmath211 , @xmath212 , and @xmath213,title="fig:",width=151,height=151 ] : @xmath211 , @xmath212 , and @xmath213,title="fig:",width=151,height=151 ] from these pictures , we can find * another fluctuation point @xmath214 . * therefore , in order to have a transition from a chaotic behavior to the regular behavior , we should cross from * two fluctuation points @xmath209 and @xmath214 . *
in this paper , we present a mathematical model of the mutation in the biological environment having 3 alleles .
we have presented two types of mutations .
we have shown that a mutation ( a mixing ) in the system can be considered as a convex combination of mendelian inheritances ( extreme or non - mixing systems ) .
the first mutation is a convex combination of two li - yorke chaotic systems and the second mutation is a convex combination of li - yorke chaotic and regular systems . in the first case ,
the first mutation can be considered as an evolution process between two different chaotic biological systems . in this case , we have shown that there is one fluctuation point ( transition point ) in order to change one chaotic system into another chaotic system . in the second case ,
the second mutation can be considered as an evolution process between chaotic and regular biological systems . in this case , we need two fluctuation points ( transition points ) in order to change the chaotic system into the regular system . we hope that we can find a similar phenomenon in nature .
ganikhodzhaev r.n . ,
mukhamedov f.m .
, rozikov u.a .
quadratic stochastic operators : results and open problems . _ infinite dimensional analysis , quantum probability and related topics_. vol . * 14*. no . 2 ( 2011 ) , 279335 . | in this short communication , we shall explore a nonlinear discrete dynamical system that naturally occurs in population systems to describe a transmission of a trait from parents to their offspring .
we consider a mendelian inheritance for a single gene with three alleles and assume that to form a new generation , each gene has a possibility to mutate , that is , to change into a gene of the other kind .
we investigate the derived models .
a numerical simulation assists us to get some clear picture about chaotic behaviors of such models .
0.3 cm _ mathematics subject classification 2010 _ : 92bxx , 37d45 , 39a33 .
+ _ key words _ : li - yorke chaos ; mutation ; regular transformation ; non - ergodic transformation ; quadratic stochastic operator . |
disordered systems are characterized by two kinds of variables : dynamical variables , with mean relaxation time @xmath12 , and `` structural '' variables , related to the randomness of the system and with mean relaxation time @xmath13 . as introduced by brout @xcite , there are two limits in which disorder problems can be formally treated , namely , the annealed and the quenched limits .
the latter is appropriate to systems where disorder relaxes in a much longer time scale than dynamical variables ( @xmath14 ) , and so disorder variables can be considered as effectively frozen .
the opposite situation , in which disorder and dynamical variables fluctuate in the same time scale ( @xmath15 ) , corresponds to the annealed limit .
since exact solutions can be found for some models , in these cases it is worthwhile using the annealed limit as an approximation to a quenched system , as it will be further discussed .
[ disorder]as shown by mazo @xcite , the free energy of a random quenched system is given by @xmath16 where @xmath17 is boltzmann s constant , @xmath18 is the absolute temperature of the system , @xmath19 is the fixed probability of occurrence of the random configuration @xmath20 , @xmath21 is the partition function of the system for that particular configuration , and the summations run over all possible configurations @xmath20 .
the second summation in eq.[fq ] is interpreted as due to an entropy of mixing of the random variables .
it is in general ignored , since the only way to measure an entropy is to change it , which is not possible in this case because the probabilities @xmath22 are rigid by assumption .
irrespective of that term , in general one finds that , for almost all thermodynamic functions obtained from the free energy by differentiation , the average value @xmath23 of the function can also be calculated as an average of the corresponding values @xmath24 over all the disorder configurations , @xmath25 when lattice models are considered , disorder is usually represented by site or bond random variables . for example , in problems of adsorption of gas molecules on a crystal surface one may want to consider sites with random potentials or random interactions between particles in different adsorption sites . for simplicity , consider a lattice of @xmath26 sites ( or bonds ) .
suppose that some random variable @xmath27 can take one of @xmath28 values @xmath29 at a given site ( or bond ) @xmath5 with probability @xmath30 , such that @xmath31 .
since that probability is independent of the values of the other random variables in the system , the probability of a particular configuration @xmath32 is given by @xmath33 where @xmath34 is the number of sites ( or bonds ) to which the value @xmath35 of the random variable is associated in that given configuration @xmath36 .
obviously the @xmath34 s obey the constraint @xmath37 it is possible to show that for this case the entropy of mixing in eq.[fq ] takes the form @xmath38 which is proportional to the volume of the system .
so , if one keeps the entropy of mixing term and tries to calculate ( in appropriate units ) the pressure of a quenched lattice gas the result is @xmath39 , \label{pqe}\ ] ] where @xmath40 is the pressure calculated for a particular configuration @xmath41 .
clearly eq.[pqe ] is a physically incorrect expression for the pressure , since in the zero density limit the first term vanishes , while the second term is positive .
furthermore , in the non - random limit ( @xmath42 for all @xmath43 ) all configurations are identical , but the entropy of mixing still contributes to the `` pressure '' .
finally the entropy of mixing diverges in the continuum limit ( @xmath44 , @xmath45 ) @xcite .
thus the correct expression for the pressure must lack the entropy of mixing , taking the form @xmath46 which is the expression used in previous studies of random lattice gases @xcite .
the need to drop out the entropy of mixing in the calculation of the pressure of a quenched fluid was also pointed out by singh and kovac @xcite on the basis of the interpretation of that term as an information entropy , introduced by sobotta and wagner @xcite . for most systems
the quenched limit is more realistic than the annealed limit , since in general @xmath47 , as it happens in spin glasses , for instance .
however , the exact mathematical treatment of the quenched limit is very difficult , due to the fact that the calculation of the free energy involves averaging logarithms of partition functions . on the other hand
, the annealed free energy can be written in the form @xmath48 where @xmath49 denotes the average of the partition function over the disorder configurations .
this is often calculated with the constraint that the thermal averages of the random variables obey some prescribed distribution , which is realized by the introduction of lagrange multipliers playing the role of pseudo - chemical potentials .
this implies that in the annealed limit the disorder variables are adjusted so as to minimize the free energy , a procedure which singles out a subset of the disorder configurations and introduces correlations between the disorder variables themselves .
so , the annealed limit is in general not a good approximation to the quenched limit .
nevertheless , morita @xcite showed that a quenched system can be represented by a fictitious ( annealed ) equilibrium system subject to an additional potential , composed of an infinite series of lagrange multipliers adjusted by the ( quenched ) @xmath50-``particle '' correlation functions ( @xmath51 ) for the random variables .
indeed , for various ising models it can be verified that by controlling correlations between pairs of random variables ( @xmath52 ) results for annealed systems approach those of the corresponding quenched systems @xcite .
this treatment introduces mathematical difficulties which are of course increasingly greater , and so this work will analyze the usual case in which one tries to control only the concentration of the random variables ( @xmath53 ) . in the next section
it will be shown that an entropy of mixing term can also be identified in a large class of annealed models , and must likewise be discarded in order to produce meaningful results .
consider a @xmath26-site @xmath11-dimensional lattice gas model whose particles interact via an arbitrary uniform ( non - random ) potential plus short - range interactions satisfying the annealed probability distribution @xmath54 for all pairs of nearest - neighbor sites @xmath43 and @xmath5 .
the hamiltonian of the model can be written as @xmath55 where @xmath56 is the uniform term , the summation runs over all nearest - neighbor pairs of sites and @xmath57 for any site @xmath43 , variable @xmath58 if the site is empty or @xmath59 if the site is occupied by a particle .
following the approach of thorpe and beeman for the ising model @xcite , this situation can be formally treated by writing @xmath60 in the form @xmath61 and introducing @xmath28 pseudo - chemical potentials @xmath62 to control the averages of the random numbers @xmath63 , which for all @xmath64 are subject to the constraints @xmath65 the annealed grand partition function of the system is given by @xmath66 \sum_{\{n\}}\exp \left [ \beta \mu \sum_{j=1}^{n}n_{j}-\beta h\right ] , \ ] ] where the primed summation runs over those configurations satisfying the first two constraints in [ vinc ] .
after performing the partial trace over the random variables one obtains @xmath67 \prod_{<i , j>}\left\ { \sum_{k=1}^{m}\exp \left [ \beta \left ( \xi _ { k}+j_{k}n_{i}n_{j}\right ) \right ] \right\ } .
\label{zanp}\ ] ] due to the fact that @xmath68 for all sites the term in braces in eq.[zanp ] can be expressed as @xmath69 \equiv ae^{kn_{i}n_{j}},\ ] ] if @xmath70 and @xmath71 are defined by the expressions @xmath72 with @xmath73 .
equation [ zanp ] can now be written as @xmath74 \equiv a^{nq/2}\mathcal{q}(k ) , \label{zak}\ ] ] where @xmath75 is the coordination number of the lattice and @xmath76 is just the grand partition function of a regular lattice gas with short - range interaction energy @xmath77 plus the uniform interaction potential .
the fugacities @xmath78 are eliminated by imposing for all nearest - neighbor pairs @xmath64 and all values of @xmath79 the condition @xmath80 which can be rewritten as @xmath81 where @xmath82 is the nearest neighbor pair correlation function for the regular lattice gas with interaction @xmath77 . by rearranging terms in eq.[qak ] and imposing the condition @xmath83 one obtains @xmath84 + 1 - 2\epsilon ( k)}=0 , \label{kek}\ ] ] which determines @xmath71 if one is able to calculate @xmath85 .
notice that in general the effective interaction @xmath77 depends on the temperature .
the density of the gas is set by the condition @xmath86 and from this equation it is possible to eliminate the fugacity @xmath87 .
the annealed free energy is given by @xcite @xmath88 , \ ] ] which after straightforward calculations can be written in the form @xmath89
-\sum_{k=1}^{m}q_{k}\ln q_{k}\right\ } , \label{fanri}\ ] ] where @xmath90\ ] ] is the free energy of the regular system .
the last summation on the right hand side of eq.[fanri ] is an entropy of mixing term , identical to that in eq.[pqe ] , and must be dropped as discussed in sec.[disorder ] . with that correction
, the free energy per lattice site takes the form @xmath91 , \label{fnoei}\ ] ] and the pressure of the system can then be calculated from the thermodynamic relation @xmath92
@xmath93 \right\ } , \label{pri}\ ] ] the first term being just the pressure of the regular lattice gas with effective interaction @xmath77 . in the non - random limit @xmath94 for all @xmath95 eq.[kek ] gives @xmath96 and according to eq.[pri ] the pressure is given by @xmath97 which is obviously the correct expression . from the results of this section
it is then clear that , as in the quenched limit , the entropy of mixing term can be made explicit in the free energy of any lattice gas with annealed short - range interactions , and that the correct equation of state can be obtained by discarding that term .
this is a fairly general result and will be applied in the next section to the @xmath0 lattice gas with random nearest - neighbor and uniform long - range interactions .
the existence of exact solutions @xcite for the uniform model [ unimod ] allows one to exactly solve the model in the presence of random annealed nearest - neighbor interactions , as shown in the previous section .
the hamiltonian of a @xmath0 lattice gas with uniform long - range interactions and random nearest - neighbor interactions is given by @xmath98 where the nearest - neighbor interactions @xmath99satisfy the annealed distribution @xmath100 the thermodynamic properties of this random model can be obtained from those of a reference system without the long - range interactions .
in particular the helmholtz free energy is given by @xcite @xmath101 , \ ] ] where @xmath102 is the free energy of the reference system as a function of the density @xmath103 and temperature @xmath18 , while `` ce '' denotes the convex envelope of the function in square brackets .
correspondingly , the pressure @xmath104 and chemical potential @xmath105 are given by @xcite @xmath106 , \quad \mu = \tilde{\mu}% -8i\rho , \label{plr}\ ] ] where `` mc '' denotes the maxwell construct of the function in square brackets and @xmath107 are the corresponding functions for the reference system .
from the results of the previous section , eqs.[fnoei ] and [ pri ] , the free energy @xmath108 and pressure @xmath109 of the reference system without long - range interaction are given in terms of the corresponding functions @xmath110 and @xmath111 for a regular model with effective nearest - neighbor interaction @xmath77 .
the grand - partition function of this regular model is given by @xmath112 , \ ] ] and by using the transfer matrix method with cyclic boundary conditions one easily finds @xmath113 ^{1/2}\right\ } , \label{pref}\ ] ] where @xmath114 .
the density of the system is then given by the relation @xmath115 \left [ \left ( 1-\tilde{z}e^{4k}\right ) ^{2}+4\tilde{z}\right ] ^{-1/2}}{1+\tilde{z}% e^{4k}+\left [ \left ( 1-\tilde{z}e^{4k}\right ) ^{2}+4\tilde{z}\right ] ^{1/2}}% \right\ } , \label{dens}\ ] ] from where the fugacity @xmath116 of the reference system can be determined , and the nearest - neighbor pair correlation function of the regular system is given by @xmath117 ^{-1/2}}{1+\tilde{z}e^{4k}+% \left [ \left ( 1-\tilde{z}e^{4k}\right ) ^{2}+4\tilde{z}\right ] ^{1/2}}% \right\ } .
\label{correl}\ ] ] finally , the pressure of the random system with long - range interactions described by the hamiltonian in eq.[hran ] is @xmath118 -4i\rho ^{2}\right\ } , \label{pran}\ ] ] with @xmath71 determined from the solution of eq.[kek ] .
so , the complete equation of state can be obtained from eqs.[pref]-[pran ] .
an alternative approach can be used to solve this problem , namely , the mapping onto the random ising model .
this is discussed in appendix [ equivalence ] and has the advantage of making evident the breaking of hole - particle symmetry .
furthermore , it provides a way of avoiding the use of the maxwell construct by previously locating free energy minima , which directly indicate phase transitions .
the results above are applied to the case in which the nearest - neighbor interactions are selected from the bimodal distribution @xmath119 where the interactions @xmath120 are assumed repulsive ( @xmath121 ) .
this is done in order to introduce competition effects , which , as in the pure model , are responsible for complex behavior . in the absence of competition between short- and long - range interactions
the system presents at most one phase transition .
the ground state properties of the model , discussed in detail in appendix [ ground ] , indicate that at @xmath10 and for @xmath122 the system undergoes at most three phase transitions . for @xmath123 , when there is competition between the nearest - neighbor interactions , at fixed pressure the system can present itself in structures characterized by @xmath7 ( gas phase ) , @xmath124 , @xmath125 ( liquid phases ) , and @xmath9 ( close - packed phase ) . for @xmath126
all nearest - neighbor interactions are repulsive , and the stable phases are those for which @xmath7 ( gas ) , @xmath127 , @xmath125 ( liquid phases ) and @xmath9 ( close - packed ) . for @xmath128
the system undergoes at most two transitions , associated with those of the pure model @xcite , the liquid - phase density being shifted to @xmath125 .
the characterization of the phases @xmath129 , @xmath124 and @xmath130 as liquid phases is based on the instability of the corresponding structures . as in the uniform case ,
the @xmath129 phase at @xmath10 is stable at certain pressures because particles are forced by nearest - neighbor repulsion to occupy every other site of the lattice . at any finite temperature ,
thermal fluctuations overcome the effect of the repulsion , inducing local fluctuations of the density and destroying the structure .
the stability of the @xmath124 and @xmath125 phases at @xmath10 , on the other hand , is related to the existence of interaction domains ( appendix [ ground ] ) .
these are a consequence of the interactions mobility in the annealed limit and are formed as particles are added to the system at fixed volume in order to minimize the free energy . in the @xmath124 phase all sites in the @xmath131 domains
are occupied , while all those in @xmath120 domains are empty .
for @xmath125 all sites in @xmath131 domains and every other site in @xmath120 domains are occupied .
however , interaction domains are also destroyed by thermal flucuations at any finite temperature , making the corresponding structures unstable . in the following subsections the properties of the model at all temperatures
are discussed for the various possible cases .
the relevant parameters are renormalized by @xmath132 , @xmath133 in passing it should be mentioned that , for repulsive long - range interaction ( @xmath134 ) , the system does not undergo a phase transition irrespective of the sign of the short - range interaction . for dilute short - range interactions
there are at most two phase transitions , whose critical temperatures are different for @xmath135 because of the breakdown of hole - particle symmetry , as already stated .
the critical temperature @xmath136 of the gas - liquid ( g - l ) transition is always higher than @xmath137 , the critical temperature of the liquid - close - packed ( l - cp ) transition .
this is in agreement to what is verified for the corresponding non - random continuum model @xcite , in which there is no hole - particle symmetry . for @xmath138
there is always a range of values of @xmath139 in which one or two triple points exist , as can be seen from the reentrant behavior of the triple point temperature ( @xmath140 ) curves in the @xmath141 diagrams of figs .
1(a)-(c ) , for @xmath142 , @xmath143 and @xmath144 , respectively .
the critical temperature of the g - l transition coincides with the triple point temperature for @xmath145 , whose dependence on @xmath146is shown in fig .
it can be seen that @xmath147 approaches @xmath148 as @xmath149 , while for @xmath150 , where there is no short - range repulsion , @xmath147 approaches @xmath151 .
this is the value of @xmath139 for which @xmath140 vanishes for all @xmath152 when @xmath128 .
examples of the behavior of the isotherms and phase diagrams for @xmath142 are shown in figs . 2 - 4 .
2 , for @xmath153 , there are two - phase transitions even at @xmath10 and there is no triple point . for low temperatures
the derivative of the l - cp transition pressure with respect to @xmath18 is negative .
this is related to the destruction by thermal fluctuations of the interaction domains built at @xmath10 , which makes the effective short - range repulsion weaker , lowering the pressure needed to pack the particles . in fig .
3 , for @xmath154 , there is only one phase transition ( g - cp ) at @xmath10 , although two transitions occur for a range of temperatures above the triple point . in fig .
4 , for @xmath155 , the critical temperature of the l - cp transition coincides with the triple point temperature , and for @xmath156 there is only one phase transition .
the behavior of the system when there are two triple points will be exemplified in the next subsection .
when there is competition between short - range interactions even a small amount of randomness is responsible for interesting behavior , as can be seen in the @xmath157 projection of the coexistence surfaces shown in figs .
5(a ) and ( b ) for @xmath158 ( @xmath159 ) , @xmath160 and two slightly different values of @xmath139 . in both cases
there are three phase transitions at @xmath10 , defining in decreasing density the close - packed , the two liquid ( l@xmath161 and l@xmath162 ) and the gas phases .
the lowest critical temperature corresponds to a l@xmath162-g transition . above
that there is an intermediate range of temperatures in which the system presents only two transitions . for @xmath163 that range
is limited by a triple point , immediately above whose temperature there is again three phase transitions , as shown in fig .
but for @xmath164 that triple point is replaced by another critical point , giving rise to a coexistence `` bubble '' , as can be seen in figs .
the @xmath141 diagram for @xmath159 and @xmath165 is shown in fig .
when the long - range attraction is sufficiently weak the system can present three phase transitions , whose critical temperatures are shown as solid lines in the figure .
triple points ( broken lines ) are present for intermediate values of @xmath139 .
the breaking and the reentrant behavior of the @xmath136 curve , evident in the inset of fig .
6 , is related to the appearance of coexistence `` bubbles '' . that unusual behavior is observed only for values of @xmath152 between @xmath166 and @xmath167 .
the system may also present a quadruple point , i.e. coexistence of four phases , if @xmath168 ( @xmath169 ) .
this is exemplified for @xmath170 and @xmath142 in the @xmath141 diagram in fig .
7 , where four triple point curves ( dotted lines ) intersect at the quadruple point @xmath171 .
figures 8 and 9 show @xmath172 and @xmath157 diagrams for @xmath170 , @xmath173 and values of @xmath139 around @xmath174 , illustrating the possible existence of two triple points or a quadruple point . as detailed in appendix b , when all short - range interactions are repulsive the possible structures at @xmath10 correspond to @xmath9 ( close - packed phase ) , @xmath125 ( liquid-1 phase ) , @xmath129 ( liquid-2 phase ) and @xmath7 ( gas phase ) .
when @xmath132 is much larger than @xmath176 the behavior of the system is expected to be similar to the dilute case ( @xmath177 ) .
this is confirmed for the case @xmath178 ( @xmath179 ) , as shown in the @xmath180 diagrams in figs .
10(a)-(c ) and the @xmath181 diagram in fig .
10(d ) . except for the presence of three phase transitions for small values of @xmath139 ,
the behavior is similar to that observed in fig .
isotherms and phase diagrams are shown in figs .
11 and 12 for @xmath144 , @xmath182 and @xmath183 , @xmath184 , respectively .
as expected for @xmath185(see appendix b ) the two - transition regime at @xmath10 consist of the @xmath7 , @xmath186 and @xmath9 phases , as in the dilute case . for small values of the long - range interactions
, the great energy difference between short - range interactions leads to the high pressure needed to close - pack the system , when it is necessary to occupy nearest neighbor sites with @xmath120 bonds . at low pressure the occupation of those sites
is virtually forbidden , giving rise to the behavior shown in figs .
11 and 12 . as the value of @xmath139
is increased , the system becomes more similar to the dilute case , as can be seen in figs .
10(a)-(c ) . when @xmath187 ( @xmath188 ) both two - transition regimes at @xmath10 , described in appendix b are possible , with the liquid - phase corresponding to @xmath129 or @xmath125 if @xmath152 is less or greater than @xmath189 , respectively . at finite temperatures the case
@xmath190 contains ingredients of both regimes .
the @xmath180 diagram for the case @xmath191 ( @xmath192 ) and @xmath193 is shown in fig . 13 .
notice that both triple point lines touch the @xmath139 axis at the same point , indicating the absence of a two - transition regime at @xmath10 .
figure 14 shows isotherms and phase diagrams for the same case and @xmath194 , where a triple point is observed , corresponding to the coexistence of the close - packed and the two liquid phases .
the l@xmath161-l@xmath162 transition line has a positive temperature derivative , as shown in fig .
14(b ) , differently from what is observed for @xmath195 , as shown in fig .
8(a ) . for @xmath196 and in the two - transition regime ,
the @xmath10 liquid - close - packed transition pressure is given by @xmath197 $ ] , the arithmetic mean of the corresponding results for the pure cases @xmath198 and @xmath199 . for @xmath200 the system can present a quadruple point , as observed for the @xmath169 case .
the helmholtz free energy of any lattice gas model with random short - range interactions in the annealed limit was shown to present an entropy of mixing term . as shown for the quenched limit
, that term has to be dropped from the free energy in order to give correct physical expressions for the pressure . as an application ,
the one - dimensional lattice gas with uniform infinite range interactions and random short - range interactions satisfying bimodal distributions was exactly solved .
explicit results were presented for the cases @xmath128 , @xmath201 , @xmath202 , @xmath203 and @xmath204 , where there is competition between short- and long - range interactions . in general , the system can present at most three phase transitions at a fixed temperature .
the phases were identified as close - packed , gas and two liquid phases , the latter being characterized at low temperatures by the existence of interaction domains . a very important effect of disorder was the breaking of the hole - particle symmetry of the pure model , which induces different critical temperatures for the transitions . under certain conditions
phase diagrams present negative thermal derivatives of some transition pressures , a phenomenon observed in nature in water , for instance .
the existence of triple and quadruple points was demonstrated , and for @xmath123 and small values of @xmath152 the existence of coexistence `` bubbles '' was observed . by using the transformation @xcite @xmath205
it can be shown that the annealed grand partition function of the random lattice gas , @xmath206 \sum_{\{n\}}\exp \left [ \beta \mu \sum_{j=1}^{n}n_{j}-\beta h\right ] , \ ] ] can be written as @xmath207 z_{\sigma } , \ ] ] where @xmath208 , @xmath209 and @xmath210 \label{zising}\ ] ] is the annealed grand partition function of a random ising model with @xmath211 and @xmath212 j_{k}.\ ] ] notice that the distributions of the random fields @xmath213 and the short - interactions @xmath27 are correlated .
the presence of the random terms in the effective field @xmath213 acting on the spin in the @xmath5th site breaks the @xmath214 @xmath215 , @xmath216 symmetry present in the uniform ising model ( for which @xmath217 for all @xmath79 ) .
the effect of this on the behavior of the gas is that , unlike the pure model , the hole - particle symmetry is also broken in the presence of any randomness , and so all transitions have different critical temperatures .
this is in contrast to the lattice gas with particles interacting via softened core uniform potentials @xcite , where multiple phase transitions are also present , but whose critical temperatures are not all different because the symmetry is still present .
the correlation between the random fields and random interactions allows one to perform the partial trace over the disorder variables , mapping the system onto a regular ising model with an effective interaction @xmath77 and a interaction dependent effective field . rewriting eq.[zising ] as @xmath218 \right\ } ,
\ ] ] where @xmath219 , \ ] ] and performing a decimation @xmath220 \right\ } \equiv ae^{k\left ( \sigma _ { j}\sigma _ { j+1}-\sigma _ { j}-\sigma _ { j+1}\right ) } , \ ] ] with @xmath221 @xmath222 and @xmath223 one can express the annealed grand - partition function @xmath224 as @xmath225 , \label{ziref}\ ] ] where @xmath226 by eliminating @xmath70 and the fugacities @xmath227 the annealed free energy of the random ising model ( as a function of the magnetization ) can be written as @xmath228e^{2(k-\beta j_{k})}+[1+\epsilon ( k)]e^{-2(k-\beta j_{k } ) } } , \label{graclfi}\end{aligned}\ ] ] in which once more there is an entropy of mixing term and @xmath229 \right\ } + i\sigma ^{2}\ ] ] is the free energy of an ising model with average magnetization @xmath230 in a uniform field and uniform short- ( @xmath77 ) and long - range ( @xmath231 ) interactions @xcite , @xmath71 being determined from the solution of the equation @xmath232 + \left [ 1-\epsilon _ { \sigma } ( k)\right ] } = 0.\ ] ] the nearest - neighbor spin - spin correlation function is explicitly given by @xmath233 by inverting the transformation [ lgim ] the annealed free energy of the lattice gas can be written as @xmath234 from where the pressure is calculated by explicitly excluding the entropy of mixing term present in @xmath235 , @xmath236 @xmath237 while the density is given by @xmath238 by studying the properties of the ising model described by the free energy @xmath239 in eq.[graclfi ] the behavior of the lattice gas can be determined .
the equilibrium state of the magnetic system at fixed temperature and external field @xmath240 is that which minimizes @xmath241 . by locating those minima and using eqs.[plg ] and [ dlg ]
the equation of state of the gas is exactly determined .
this procedure automatically removes the unstable portions of the isotherms and is equivalent to the maxwell construct .
[ ground]at zero temperature the helmholtz free energy @xmath242 of a system is identical to its internal energy , as it is evident from the thermodynamic relation @xmath243 where @xmath18 is the temperature and @xmath244 is the entropy .
the ground state behavior of the system can thus be determined by writing the internal energy @xmath245 for the various ranges of the density @xmath103 and calculating the pressure using the expression @xmath246 , \ ] ] where `` mc '' denotes the maxwell construct @xcite of the function in square brackets , which substitutes horizontal parts ( corresponding to phase transitions ) for the unstable portions of the isotherms .
a fundamental feature of annealed models is that impurities are free to rearrange themselves so as to minimize the free energy . for a lattice gas with random interactions
this means that , as particles are added to the system , interactions of the same kind have a tendency to aggregate , forming domains . from this analysis it is possible to find those structures which minimize the internal energy and determine the ground state behavior of the model discussed in sec .
[ slrint ] , which is summarized below . in the following discussion
the volume of the system is regarded as fixed .
* case * @xmath247 . in this situation , because of the presence of the attractive interactions @xmath131 , interaction domains are formed even at infinitesimal densities . for sufficiently weak long - range attraction energy @xmath231 , the density can be increased from @xmath7 to @xmath248 at zero pressure by adding particles to the @xmath131 domains . at densities between @xmath124 and @xmath125 new particles can be allocated with minimum energy cost at every other site of the @xmath120 domains .
the pressure is nevertheless raised , because of the smaller volume available . finally , for @xmath249 particles can be added only by allocating them in empty sites between two other particles in the @xmath120 domain , raising both internal energy and pressure . for strong long - range attraction
some of the transitions are suppressed , the various possibilities for the @xmath10 isotherms being as follows .
@xmath250 - 2(1+p)i\equiv p_{3 } \end{array } \right .
\vspace{0.15 cm } \\
\mbox{condition : } i<\min \left\ { -\frac{2}{1+p}j_{b},\frac{2}{1+p}% j_{a}\right\ } \end{array } \right . \\ \\ \left\ { \begin{array}{c } \mbox{\textbf{2 transitions:}}\vspace{0.15 cm } \\ \left\ { \begin{array}{l }
0<\rho < \frac{1}{2}(1+p),\quad p = p_{1}\vspace{0.15 cm } \\
\frac{1}{2}(1+p)<\rho < 1,\quad p = p_{3 } \end{array } \right . \vspace{0.15 cm } \\ \mbox{conditions : } \left\ { \begin{array}{l } p>\frac{j_{a}+j_{b}}{j_{a}-j_{b}}\equiv p_{d}\quad \mbox{and}\vspace{0.15 cm } \\
\frac{2}{1+p}j_{a}<i<2\left [ \frac{p}{1+p}j_{a}-j_{b}\right ] \end{array } \right . \end{array } \right .
\\ \left\ { \begin{array}{c } \mbox{\textbf{2 transitions:}}\vspace{0.15 cm } \\
\left\ { \begin{array}{l } 0<\rho < p,\quad p = p_{1}\vspace{0.15 cm } \\ p<\rho < 1,\quad p=4p(j_{a}-j_{b}-i)=p_{2}^{\prime } \end{array } \right . \vspace{0.15 cm }
\\ \mbox{conditions : } \left\ { \begin{array}{l } j_{a}>-j_{b}\quad \mbox{and}\vspace{0.15 cm } \\
p < p_{d}\quad \mbox{and}\vspace{0.15 cm } \\
-\frac{2}{1-p}j_{b}<i < j_{a}-j_{b } \end{array } \right . \end{array } \right . \\ \\
\left\ { \begin{array}{c } \mbox{\textbf{1 transition : } } \vspace{0.15 cm } \\
\left\ { 0<\rho < 1,\quad p = p_{1}\right .
\vspace{0.15 cm } \\ \mbox{conditions : } \left\ { \begin{array}{l } p < p_{d}\quad \mbox{and}\quad i > j_{a}-j_{b}\vspace{0.15 cm } \\ \mbox{or}\vspace{0.15 cm } \\
p > p_{d}\quad \mbox{and}\quad i>2\left [ \frac{p}{1+p}j_{a}-j_{b}\right ] \end{array } \right .
\end{array } \right . \end{array}\ ] ] * case * @xmath175 this is the case where all nearest - neighbor interactions are repulsive . for sufficiently weak long - range interactions
@xmath231 there is no aggregation of particles at low densities . on the contrary
, particles can be added to the system at minimum cost by allocating them such that they have no nearest neighbors . in general this
can be done at zero pressure only if @xmath251 . for greater densities interaction domains
are formed , and new particles are allocated , by raising the pressure , between occupied sites in the @xmath131 domains , which are filled up for @xmath252 .
adding further particles requires even higher pressures , since they must occupy available sites in the @xmath253domains .
again the existence of the various possible transitions depends on the strength of the long - range interactions .
zero temperature isotherms are described below .
@xmath254 - 2(1+p)i\equiv p_{3 } \end{array } \right .
\vspace{0.15 cm } \\
\mbox{condition : } i<\min \left\ { -\frac{4}{1+p}j_{a},4(j_{a}-j_{b})\right\ } \end{array } \right .
\left\ { \begin{array}{c } \mbox{\textbf{2 transitions : } } \vspace{0.15 cm } \\
\left\ { \begin{array}{l } 0<\rho < \frac{1}{2}(1+p),\quad p = p_{1}\vspace{0.15 cm } \\ \frac{1}{2}(1+p)<\rho < 1,\quad p = p_{3 } \end{array } \right .
\vspace{0.15 cm } \\ \mbox{conditions : } \left\ { \begin{array}{l } j_{a}>\frac{2}{3}j_{b}\quad \mbox{and}\vspace{0.15 cm } \\ p>\frac{2j_{a}-j_{b}}{j_{a}-j_{b}}\equiv p_{s}\quad \mbox{and}\vspace{0.15 cm } \\ -\frac{4}{1+p}j_{a}<i<2\left [ \frac{p}{1+p}j_{a}-j_{b}\right ] \end{array } \right . \end{array } \right .
\left\ { \begin{array}{c } \mbox{\textbf{2 transitions : } } \vspace{0.15 cm } \\
\left\ { \begin{array}{l } 0<\rho < \frac{1}{2},\quad p = p_{1}\vspace{0.15 cm } \\
\frac{1}{2}<\rho < 1,\quad p=-4[pj_{a}+(1-p)j_{b}+i]=p_{2}^{\prime } \end{array } \right . \vspace{0.15 cm } \\ \mbox{conditions : } \left\ { \begin{array}{l } j_{a}<\frac{1}{2}j_{b}\quad \mbox{and}\vspace{0.15 cm } \\
p < p_{s}\quad \mbox{and}\vspace{0.15 cm } \\ 4(j_{a}-j_{b})<i<-2[pj_{a}+(1-p)j_{b } ] \end{array } \right .
\end{array } \right . \\ \\ \left\ { \begin{array}{c } \mbox{\textbf{1 transition : } } \vspace{0.15 cm } \\ \left\ { 0<\rho < 1,\quad p = p_{1}\right .
\vspace{0.15 cm } \\
\mbox{conditions : } \left\ { \begin{array}{l } p < p_{s}\quad \mbox{and}\quad i>-2[pj_{a}+(1-p)j_{b}]\vspace{0.15 cm } \\ \mbox{or}\vspace{0.15 cm } \\
p > p_{s}\quad \mbox{and}\quad i>2\left [ \frac{p}{1+p}j_{a}-j_{b}\right ] \end{array } \right . \end{array } \right . \end{array}\ ] ] * figure 1 . * ( a - c ) @xmath180 diagram , where @xmath255 is the renormalized temperature and @xmath256 , for dilute short - range interactions ( @xmath128 ) with probabities @xmath142 , @xmath143 and @xmath144 , respectively .
continuous curves represent critical temperatures ( @xmath257 and @xmath258 ) of the phase transitions , while triple point temperatures ( @xmath259 ) are shown as dotted curves .
curves for @xmath260 and @xmath259 coincide at point @xmath261 , whose @xmath262 coordinate ( @xmath147 ) is shown in ( d ) as a function of @xmath152 .
* figure 2 .
* ( a ) isotherms @xmath263 ( continuous curves ) , where @xmath264 is the renormalized pressure , and coexistence curves ( dotted curves ) for dilute short - range interactions ( @xmath128 ) , @xmath265and @xmath153 ; ( b ) phase diagram @xmath266 showing close - packed ( cp ) , liquid ( l ) and gas ( g ) phases .
continuous curves correspond to first - order transitions terminating at the critical temperatures @xmath267 and @xmath258 .
* figure 4 . * ( a ) the same as in fig .
2(a ) for @xmath268 ; ( b ) phase diagram showing the coincidence of the triple point ( temperature @xmath259 ) and the critical point of the liquid - close - packed ( temperature @xmath258 ) .
* figure 5 . * @xmath269 projections of the coexistence surface for @xmath159 ( @xmath270 ) , @xmath160 and ( a ) @xmath163 and ( b ) @xmath164 . the triple point present in ( a ) , whose temperature is @xmath259 ,
is suppressed in ( b ) , where a coexistence `` bubble '' can be observed , as indicated by the arrows .
* figure 6 .
* @xmath180 diagram for @xmath159 and @xmath160 .
the inset shows the breaking and reentrant behavior of the @xmath271 curve , as well as the existence of the new points @xmath272 and @xmath273 , extremes of a new triple point line ( dotted ) .
the @xmath258 curve is ommited from the inset for clarity .
* figure 8 .
* phase diagram for @xmath170 , @xmath173 and ( a ) @xmath274 , ( b ) @xmath275 and ( c ) @xmath276 .
two triple points are evident in ( a ) and ( c ) , while the quadruple point is present in ( b ) . in all cases
the axes intervals are @xmath277 and @xmath278 .
* figure 9 . * @xmath269 ( @xmath279 ) projections of the coexistence surfaces for @xmath280 , @xmath173 and ( a ) @xmath274 , ( b ) @xmath275 and ( c ) @xmath276 .
the broken lines indicate the triple point temperatures in ( a ) and ( c ) , and the quadruple point temperature in ( b ) .
* figure 10 . *
( a - c ) @xmath180 diagram , where @xmath255 is the renormalized temperature and @xmath256 , for the case @xmath203 ( @xmath281 ) with probabities @xmath143 , @xmath143 and @xmath282 , respectively .
continuous curves represent critical temperatures ( @xmath271 and @xmath258 ) of the phase transitions , while triple point temperatures ( @xmath259 ) are shown as dotted curves .
coordinates @xmath147 and @xmath283 of points @xmath261 and @xmath284 , as well as values @xmath285 and @xmath286 at which triple point temperatures become zero , are shown in ( d ) as a function of @xmath152 .
notice that , while @xmath147 goes to @xmath287 as @xmath149 , @xmath283 goes to @xmath288 as @xmath150 , as expected .
* figure 11 . *
( a ) @xmath263 isotherms ( solid lines ) and coexistence curves ( dotted lines ) for @xmath289 , @xmath144 and @xmath182 ; ( b ) @xmath290 phase diagrams showing the various transition lines and critical temperatures ; ( c ) @xmath269 projections of the coexistence curves shown in ( a ) .
* figure 14 . * ( a ) @xmath263 isotherms ( solid lines ) and coexistence curves ( dotted lines ) for @xmath293 , @xmath144 and @xmath194 ; ( b ) @xmath290 phase diagrams showing the various transition lines , critical temperatures and the triple point . | an analysis of the random lattice gas in the annealed limit is presented .
the statistical mechanics of disordered lattice systems is briefly reviewed .
for the case of the lattice gas with an arbitrary uniform interaction potential and random short - range interactions the annealed limit is discussed in detail . by identifying and extracting an entropy of mixing term , a correct physical expression for the pressure
is explicitly given .
the one - dimensional lattice gas with uniform long - range interactions and random short - range interactions satisfying a bimodal annealed probability distribution is discussed .
the model is exactly solved and is shown to present interesting behavior in the presence of competition between interactions , such as the presence of three phase transitions with different critical temperatures and the occurrence of triple and quadruple points .
[ intro]the lattice gas model @xcite was introduced by lee and yang in 1952 as an application of their theory of condensation , due to its equivalence to the ising model . despite its somewhat artificial character
, it has been extensively and successfully applied to the study of both solid - liquid and liquid - gas transitions @xcite .
it has also been applied to problems of adsorption of a gas on a crystal surface @xcite .
the mapping of the two - dimensional lattice gas with nearest - neighbor interaction into the ising model in the square lattice allowed the exact calculation of the coexistence curve of a fluid transition exhibited by the lattice gas at sufficiently low temperatures .
however , the corresponding equation of state could not be exactly determined , due to the absence of an exact solution for the ising model in the presence of an external magnetic field .
such a solution exists for the one - dimensional analog of the model , although it undergoes no phase transition at finite temperature , having very little to contribute to even a qualitative analysis of the behavior of real substances .
notwithstanding , in the presence of uniform infinite - range interactions phase transitions are induced even in one - dimensional systems @xcite .
furthermore , as pointed out by hemmer and stell @xcite , for lattice and continuous fluids in which the potential has both a long - range attraction and a short - range repulsion , plus a hard core , up to two phase transitions are expected .
this is indeed verified for the @xmath0 lattice gas with nearest - neighbor repulsion and long - range attraction @xcite , described by the hamiltonian @xmath1 where @xmath2 , @xmath3 and the occupation numbers @xmath4 represent the absence or presence of a particle in the @xmath5th site along the chain .
such model is mathematically equivalent to the ising chain with short- and long - range interactions in an external field @xcite , which in turn is equivalent to a linear chain approximation for higher dimensional models .
depending on the ratio @xmath6of the interaction parameters , the system can present one or two phase transitions , and in the latter case a triple point may be present @xcite .
the three possible phases have zero - temperature densities given by @xmath7 ( gas phase ) , @xmath8 ( liquid phase ) and @xmath9 ( close - packed phase ) .
however , the symmetry between unoccupied and occupied cells ( hole - particle symmetry ) @xcite makes the critical temperatures for both transitions the same , differently from what happens in the analogous continuous model @xcite . on the other hand , for random versions of the model , by looking at ground state properties one can see that the presence of disorder in the nearest - neighbor interactions induces both the breaking of the hole - particle symmetry and the appearance of additional phase transitions at @xmath10 .
due to the presence of the long - range attraction those new transitions should persist at low but finite temperatures .
it is then worthwhile examining disordered versions of the model and look for what new interesting properties may arise .
the main purpose of this paper is to study the phase diagram of the @xmath0 lattice gas with random annealed nearest - neighbor interactions and uniform long - range interactions .
this particular disordered version of the model [ unimod ] has the advantage of being exactly soluble .
although of difficult physical realization , it might be useful in studying adsorption problems , if one wishes to consider situations in which the presence of adsorbed particles induces diffusion of crystalline atoms in or near imperfect surfaces .
the outline of this paper is as follows . in order to clarify one subtle aspect of the mathematical treatment of annealed models , a preceding discussion is presented .
section [ disorder ] discusses some theoretical aspects of the statistical mechanics of disordered systems .
the @xmath11-dimensional lattice gas with an arbitrary uniform interaction potential plus random short - range interactions is treated in sec .
[ lgrsrint ] , following the approach of thorpe and beeman for the ising model @xcite .
the calculation of thermodynamic functions for this class of systems is presented , and a general method for calculating the equation of state is proposed .
then the one - dimensional lattice gas with random nearest - neigbor interactions and uniform long - range interactions is considered in sec .
[ slrint ] .
the main features of the behavior of the model are presented and discussed when the nearest - neighbor interactions are selected from a bimodal distribution . in the final section the conclusions and main results of the paper
are summarized .
the equivalence between the model and the random ising model is presented in appendix a , and in appendix b the ground state properties of the model are discussed . |
parity - time ( @xmath0 ) symmetric quantum system may possess entirely real spectrum although being non - hermitian bender98,bender02,am02,mznojil06,mznojil08,moiseyev , jin09,joglekar10,gong10,joglekar12,gong13,schen , wang .
@xmath0 symmetric system is invariant under the combined @xmath3 operator in the presence of balanced gain and loss . in the past decade
, @xmath0-symmetric system has attracted tremendous interests as it possesses unintuitive but intriguing implications . due to the similarity between the paraxial wave equation describing spatial light wave propagation and the temporal schrdinger equation for quantum system , the complex refractive index distribution satisfying @xmath4 mimics @xmath5-symmetric potentials @xmath6 , @xmath0-symmetric systems are proposed and realized in coupled optical waveguides through index guiding and a inclusion of balanced gain and loss regions @xcite .
a number of novel and non - trivial phenomena are found , such as power oscillation @xcite , coherent perfect absorbers @xcite , nonreciprocal light propagation @xcite in coupled waveguides , and recently the @xmath0-symmetric microcavity lasing @xcite and gain induced large optical nonlinear @xcite in coupled resonators .
the spectral singularity aliprl2009,unidirectionalss , alipra2011,sslonghi , ssahmed , aliprl2013,ssali and invisibility lin , feng , longhiinvisibility10,longhiinvisibility14,aliinvisibility13,aliinvisibility14,aliinvisibility15 in @xmath0-symmetric system are hot topics , where reciprocal transmission and unidirectional reflectionless in @xmath0-symmetric metamaterial are intriguing features for novel optical devises .
these devices are useful for light transport , control and manipulation kottos2012r , kottos2012,reciprocalahmed , reciprocalali .
the symmetric scattering properties are usually attributed to certain internal symmetry of a scattering centre . for instance , the parity ( @xmath1 ) symmetry , or time - reversal ( @xmath2 ) symmetry of a scattering centre leads to symmetric reflection and transmission @xcite ( @xmath2-symmetric system without unequal tunnelling amplitude is hermitian , otherwise , only reciprocal reflection or transmission holds @xcite ) . here
, we report reciprocal reflection , similar as reciprocal transmission , are both related to the @xmath0 symmetry of a scattering centre : the axial ( refection ) @xmath0 symmetry , with respect to the input and output channels , induces reciprocal reflection ( transmission ) .
recent efforts on photonic aharonov - bohm effect enable photons behaving like electrons in magnetic field .
effective magnetic field for photons can be introduced in coupled waveguides by bending the waveguides @xcite , periodically modulating the refractive index @xcite , and the photon - phonon interactions @xcite ; or in coupled resonators by magneto - optical effect @xcite , dynamic modulation @xcite , and off - resonance coupling paths imbalance @xcite . in this work ,
we focus on the @xmath0-symmetric structure with balanced gain and loss threading by synthetic magnetic flux , where photons feel a nonreciprocal tunnelling phase between neighbour resonators .
the nonreciprocal tunnellings and balanced gain and loss break the @xmath1 and @xmath2 symmetry but keep the @xmath0 symmetry of the scattering centre . the axial ( reflection )
@xmath0 symmetry will lead to reciprocal reflection ( transmission ) and unidirectional transmission ( reflection ) .
our findings provide new insights of @xmath0 symmetry and the symmetric / asymmetric scattering , which are instrumental for the applications of @xmath0-symmetric metamaterial for light transport and one - way optical devises .
* reciprocal and unidirectional scattering of @xmath0-symmetric structures . * the symmetric scattering properties of a @xmath7-symmetric structure are closely related to the classification of @xmath3-symmetry .
the parity operator @xmath1 is the spatial reflection operator , @xmath2 is the time - reversal operator . in fig .
[ fig1]a , b , we schematically show two types of @xmath8 symmetry .
the hamiltonian of the scattering centre is @xmath0-invariant , i.e. , @xmath9 .
the input and output leads are connected to the @xmath0-symmetric scattering centre at sites @xmath10 and @xmath11 . if the connection sites under the parity operation satisfies @xmath12 , @xmath13 , the system is called axial @xmath5 symmetric ( fig .
[ fig1]a ) .
if the connection sites under the parity operation satisfies @xmath14 , @xmath15 , the system is called reflection @xmath0 symmetric ( fig . [ fig1]b ) .
the red plane indicates the up - to - down ( left - to - right ) spatial reflection correspondence of axial ( reflection ) @xmath5 symmetry . in order to address the reciprocal and unidirectional scattering behavior .
we study the reflection and transmission of a scattering centre for the left side and right side inputs , respectively .
we denote the two scattering states as @xmath16 and @xmath17 for the input with wave vector @xmath18 .
the forward going and backward going waves are in form of @xmath19 . combining with the reflection and transmission coefficient , we assume the scattering state wave function ( not at the spectral singularities , see methods ) of left side input as , @xmath20where @xmath21 and @xmath22 represent the reflection and transmission coefficients for the left side input with wave vector @xmath23 .
similarly , the wave function of right side input is in form of@xmath24where @xmath25 and @xmath26 represent the reflection and transmission coefficients for the right side input with wave vector @xmath23 . *
reciprocal reflection under axial @xmath0 symmetry . * as shown in fig .
[ fig1]a , this type of @xmath0-symmetric scattering centres have connection sites under parity operation satisfying @xmath27 , @xmath13 . in the axial @xmath0-symmetric configuration ,
the @xmath0 symmetry is defined as @xmath28 , @xmath29 in the leads , and as @xmath30 in the centre .
the whole scattering system is axial @xmath5-symmetric with respect to the leads .
the axial @xmath0 symmetry results in symmetric relations on the scattering coefficients as ( see supplementary information note 1 for details ) , @xmath31from equations ( [ axialpt1],[axialpt2 ] ) , we notice the reflection probabilities for the left and right side inputs are the same , i.e.,@xmath32 in other words , the axial @xmath0 symmetry leads to the reciprocal reflection .
notice that we have reciprocal reflection @xmath33 at any transmission zero @xmath34 , where one - way pass through is possible .
furthermore , considering the waves with vectors @xmath23 and @xmath35 , the reflection and transmission coefficients further satisfy @xmath36 , @xmath37 , @xmath38 , and @xmath39 . * reciprocal transmission under reflection @xmath0 symmetry . * as shown in fig .
[ fig1]b , this type of @xmath0-symmetric scattering centres have connection sites under parity operation satisfying @xmath14 , @xmath40 . in the reflection @xmath0-symmetric configuration , the @xmath5 symmetry is defined as @xmath41 , @xmath42 in the leads , and as @xmath43 in the centre .
the whole scattering system is reflection @xmath0-symmetric with respect to the leads .
the reflection @xmath0 symmetry results in symmetric relations on the scattering coefficients as ( see supplementary information note 2 for details ) , @xmath44from equations ( [ reflectionpt1 ] , [ reflectionpt2 ] ) , we notice the transmission probabilities for the left side and right side inputs are the same , i.e. , @xmath45this indicates the reflection @xmath0 symmetry leads to the reciprocal transmission , as observed in bragg gratings and other @xmath5-symmetric structures reciprocalahmed , reciprocalali , kottos2012 .
notice that we have reciprocal transmission @xmath46 at any reflection zero @xmath47 , where unidirectional reflectionless is possible lin , feng , longhiinvisibility10,longhiinvisibility14,aliinvisibility13,aliinvisibility14,aliinvisibility15 .
furthermore , considering the waves with vectors @xmath23 and @xmath35 , the reflection and transmission coefficients further satisfy @xmath48 , @xmath49 , @xmath50 , and @xmath51 .
we show that in the present of @xmath0 symmetry , the reciprocal reflection or transmission in a scattering centre is protected when the axial or reflection @xmath0 symmetry holds even though the @xmath1 and @xmath2 symmetry are absent .
moreover , @xmath0 symmetry structure may exhibit unidirectional scattering behavior . * @xmath0-symmetric rhombic ring structures .
* we use a rhombic ring structure ( fig . [ fig1]c , d ) to elucidate the results . the scattering centre encloses with an effective magnetic flux @xmath52
, photons moving along the rhombic ring structure in clockwise ( counterclockwise ) direction will acquire an additional direction - dependent phase factor @xmath53 ( @xmath54 ) , thus photons tunnelling is nonreciprocal except when @xmath55 , @xmath56 .
this is an effective photon aharonov - bohm effect creating by synthetic magnetic field abcaging , dynmoduprl , photphon , moeffect , dynmodu , chongyd , hafezi2014 .
the phase factor @xmath57 is an analytical function of @xmath58 with period of @xmath59 , it is sufficient to understand the influence of magnetic flux on the scattering by studying @xmath58 in the region @xmath60 . to realize a synthetic magnetic field ,
two ring resonators are coupled through an auxiliary off - resonant ring resonator .
the auxiliary resonator introduces optical paths imbalance when coupling to two primary resonators , the auxiliary resonator can be effectively reduced and create a coupling phase factor between two primary resonators .
the coupled resonators under synthesized magnetic field is described by a magnetic tight - binding hamiltonian chongyd , hafezi2014 , @xmath61where @xmath62 is a nonreciprocal phase shift induced by the magnetic flux in the tunnelling constant . in fig .
[ fig1]c , the hamiltonian of the scattering centre is @xmath63 , where @xmath64 is the gain / loss rate .
the balanced gain and loss are the origin of the non - hermiticity realized in the optical systems @xcite .
the configuration is axial @xmath0-symmetric with the parity operator acting on the rhombic ring sites defined as @xmath65 , @xmath66 , @xmath67,@xmath68 . in fig .
[ fig1]d , the hamiltonian of the scattering centre is @xmath69 , the parity operator @xmath1 is defined as @xmath70 , @xmath71 , @xmath72 , @xmath73 , and the configuration is reflection @xmath0-symmetric . in the system ,
the magnetic flux is inverted meanwhile the gain and loss are switched under the @xmath1 or @xmath2 operation .
however , the system is invariant under the combined @xmath0 operator , i.e. , the presence of non - trivial magnetic flux as well as balanced gain and loss both break the @xmath1 and @xmath2 symmetry but keep the @xmath0 symmetry of the system .
the scattering centre is actually a two - arm aharonov - bohm interferometer .
light wave propagates through two pathways ( @xmath74 and @xmath75 ) between the connection sites @xmath76 , @xmath77 and interfere with each other .
the interference generates the output which varies as the enclosed magnetic flux .
the effective magnetic field is gauge invariant and the magnetic flux acts globally , thus the reflection and transmission are not affected by the nonreciprocal phase distribution in the tunnellings for fixed magnetic flux . in the following ,
we discuss the reflection and transmission of the @xmath0-symmetric rhombic ring structures in details . the reflection and transmission coefficients for the axial @xmath0-symmetric rhombic ring structure ( fig .
[ fig1]c ) are calculated from the schrdinger equations ( see methods ) , yielding @xmath78}{4e^{2ik}\cos ^{2}(\phi
/2)+4\sin ^{2}k-\gamma ^{2 } } , \label{axialptrlr } \\
t_{\mathrm{l } } & = & \frac{4i\sin k[2\cos k\cos ( \phi /2)-\gamma \sin ( \phi /2)]% } { 4e^{2ik}\cos ^{2}(\phi /2)+4\sin ^{2}k-\gamma ^{2 } } , \label{axialpttl } \\ t_{\mathrm{r } } & = & \frac{4i\sin k[2\cos k\cos ( \phi /2)+\gamma \sin ( \phi /2)]% } { 4e^{2ik}\cos ^{2}(\phi /2)+4\sin ^{2}k-\gamma ^{2}}.
\label{axialpttr}\end{aligned}\]]the reflection and transmission probabilities are functions of the magnetic flux @xmath79 , gain / loss rate @xmath64 , and wave vector @xmath23 .
they satisfy @xmath80 ( see fig . [ fig2]a , d ) , @xmath81 ( see fig .
[ fig2]b , c ) , and @xmath82 ( see fig .
fig2e , f ) . in the absence of non - trivial magnetic flux @xmath58 , or gain / loss @xmath83 ,
the system is @xmath1-symmetric ( reflection-@xmath1-symmetric for @xmath58 absence , i.e. , left to right by mirror imaging ; inversion-@xmath1-symmetric for @xmath64 absence , i.e. , left to right by @xmath84 rotation ) , the reflection and transmission are both reciprocal .
the non - trivial magnetic flux @xmath58 together with balanced gain and loss @xmath64 break the @xmath1 symmetry .
the symmetric transmission in the @xmath0-symmetric system at @xmath85 is broken , i.e. , the transmission is unidirectional at @xmath85 . moreover
, the axial @xmath0 symmetry protects the symmetric reflection , therefore , the reflection is reciprocal but the transmission is unidirectional .
the white curves in fig .
[ fig2 ] show the reflection / transmission zeros . at @xmath86
, we have @xmath87 or @xmath88 with total reflection @xmath89 .
this indicates that we only have a non - zero transmission for the right side or left side input , thus the axial @xmath0-symmetric rhombic ring structure allows one - way pass through .
photons circle in the scattering centre either in a clockwise direction or in a counterclockwise direction , we schematically illustrate the two pathways in the supplementary ( fig . [ figs1 ] ) .
the phase difference between two pathways affects the interference in the scattering centre , thus the transmission varies as the effective magnetic flux induced phase difference .
the phase difference between clockwise direction and counterclockwise direction of transmission pathways is @xmath52 for the left side input ( supplementary , see fig .
[ figs1]c ) or @xmath90 for the right side input ( supplementary , see fig .
[ figs1]d ) .
the transmission pathways are not equivalent in the presence of gain / loss , the interference of phase difference being @xmath52 is different from the interference of phase difference being @xmath90 .
therefore , the unidirectional transmission is enabled in the presence of nonreciprocal tunneling phase factor ( @xmath91 ) attributed to non - trivial magnetic flux ( @xmath92 , @xmath56 ) . in fig .
[ fig3 ] , we plot the reflection and transmission probabilities for an axial @xmath0-symmetric rhombic ring structure ( fig . [ fig1]c ) at several set parameters .
figure [ fig3]a is for a system with balanced gain and loss in the absence of magnetic flux , i.e. , @xmath93 , @xmath94 . the gain / loss , being non - hermitian , plays the role of on - site potentials and is the origin of unidirectional behavior .
however , the presence of balanced gain and loss alone does not ensure unidirectional scattering .
we notice the reflection and transmission in fig .
[ fig3]a are both reciprocal .
the scattering is unitary even though the system is non - hermitian ( the balanced gain and loss of this rhombic ring structure @xmath95 can be reduced to an anti - hermitian interaction @xmath96 by composing @xmath97 , and the non - hermiticity of the scattering centre only arises from the anti - hermitian interaction between @xmath98 , which is proved to have unitary scattering unitaryscatjl ) . by introducing magnetic flux to the system ,
the @xmath1 and @xmath2 symmetry is destroyed but the @xmath0 symmetry holds . the interference between light waves from the loss arm and the gain arm generates unidirectional transmission for non - trivial magnetic flux .
figure [ fig3]b is for a system in the presence of non - trivial magnetic flux , i.e. , @xmath93 , @xmath99 .
the unidirectional transmission zero happens at @xmath100 , i.e. , at @xmath101 , @xmath102 , @xmath103 ; at @xmath104 , @xmath105 , @xmath106 , which indicates a one - way pass through .
figure [ fig3]c is for a hermitian scattering centre in the presence of non - trivial magnetic flux , i.e. , @xmath107 , @xmath108 , we have hermitian scattering without unidirectional behavior . in the rhombic ring structure under axial @xmath0 symmetry ( fig .
[ fig1]c ) , the reflection and transmission coefficients @xmath21 , @xmath25 , @xmath22 , @xmath26 diverge at the spectral singularities aliprl2009 .
when @xmath109 , we have the reflection and transmission coefficients @xmath110/[4\sin ^{2}(\phi /2)-\gamma ^{2}]$ ] and @xmath111 $ ] .
we notice the spectral singularities are at @xmath112 .
when @xmath113 , we have the reflection and transmission coefficients @xmath114 and @xmath115 .
the spectral singularities are at @xmath116 . at the spectral singularities , the scattering states are in form of a self - sustained emission @xmath117 , @xmath118 and a reflectionless absorption @xmath119 , @xmath120 ( see method )
the transfer matrix of the scattering centre is @xmath121 , @xmath122 , @xmath123 with matrix - element @xmath124 vanishes .
now , we turn to discuss the rhombic ring structure under reflection @xmath0 symmetry .
the configuration is shown in fig .
[ fig1]d . in supplementary
, figure [ figs2 ] schematically illustrates the pathways of photons .
the connection sites are linked by two same pathways . in the presence of magnetic flux
@xmath52 , photons travelling from left lead to right lead in clockwise direction and counterclockwise direction acquire additional phases @xmath125 and @xmath126 in the two pathways ( supplementary , see fig .
[ figs2]c ) , respectively .
the situation is unchanged for photons travelling inversely from right lead to left lead ( supplementary , see fig .
[ figs2]d ) .
equivalently , the upper and lower pathways are undistinguishable .
therefore , only relative phase difference @xmath52 matters ( affecting the transmission coefficient ) and the transmission is directionless .
the reflection and transmission coefficients are calculated as ( see methods ) @xmath127 } , \label{reflectionptrl } \\
r_{\mathrm{r } } & = & \frac{(\gamma ^{2}-2\gamma \sin k-1)\cos ^{2}k+\sin ^{2}(\phi /2)}{\sin ^{2}k - e^{2ik}[\gamma ^{2}\cos ^{2}k-\cos ^{2}(\phi /2 ) ] } , \label{reflectionptrr } \\
t_{\mathrm{l } } & = & t_{\mathrm{r}}=\frac{i\sin \left ( 2k\right ) \cos ( \phi /2)% } { \sin ^{2}k - e^{2ik}[\gamma ^{2}\cos ^{2}k-\cos ^{2}(\phi /2)]}. \label{reflectionpttlr}\end{aligned}\]]the reflection and transmission coefficients are functions of the magnetic flux @xmath79 , gain / loss rate @xmath64 , and wave vector @xmath23 .
they satisfy @xmath128 and @xmath129 .
figure [ fig4 ] implies a reciprocal transmission ( fig .
[ fig4]a , d ) and unidirectional reflection ( fig .
[ fig4]b , c , e , f ) . in this configuration , the scattering with both reflection and transmission being reciprocal happens in the absence of gain and loss ( @xmath107 ) , that is when the system is @xmath1-symmetric . in the presence of gain and loss ( @xmath130 ) ,
the reflection probability is unidirectional , but the reflection @xmath5 symmetry protects the reciprocal transmission . due to the presence of gain and loss , the probability of the total reflection and transmission after scattering is not unitary , being balanced gain and loss rate dependent .
the white curves in fig .
[ fig4 ] show the reflection and transmission zeros . at @xmath131 and @xmath94 , @xmath132 , @xmath133 . at the reflection zeros
, @xmath134 , @xmath135 , and @xmath136 , the system exhibits unidirectional reflectionless with reciprocal transmission . in fig .
[ fig5 ] , we plot the reflection and transmission probabilities for a reflection @xmath0-symmetric rhombic ring structure ( fig .
[ fig1]d ) at several set parameters .
figure [ fig5]a , b are for balanced gain and loss rate @xmath93 with two different magnetic flux @xmath94 and @xmath137 , respectively . the reciprocal transmission and unidirectional reflection are clearly seen . in fig .
[ fig5]a , the unidirectional reflectionless happens at @xmath138 , @xmath139 , @xmath46 . in fig .
[ fig5]b , the unidirectional reflectionless happens at @xmath140 , @xmath141 , @xmath46 ; or at @xmath142 , @xmath143 , latexmath:[$% and loss @xmath107 , the reflection and axial @xmath0-symmetric rhombic ring configurations reduce to an identical system . in fig . [ fig5]c , we plots the reflection and transmission probabilities of a scattering centre in the absence of both gain and loss and magnetic flux , i.e. , @xmath107 , @xmath94 , we observe hermitian scattering behavior of reciprocal reflection and transmission similar as @xmath145 , @xmath108 shown in fig .
notice that no spectral singularity emerges in the scattering of reflection @xmath0-symmetric rhombic ring system .
the system with @xmath94 leads to input with wave vector @xmath109 both sides invisible that @xmath146 , @xmath147 ( black crosses in fig .
[ fig4]a - c ) ; the system with @xmath148 leads to input with vector @xmath149 both sides opaque that @xmath150 , @xmath151 . for the input with wave vector @xmath109 ,
the scattering behavior is very sensitive to the magnetic flux .
we investigate the reciprocal and unidirectional scattering of @xmath0-symmetric structures .
we show an insightful understanding of the symmetric scattering behavior , that is associated with the type of @xmath0 symmetry , defined as the @xmath0 symmetry of the connection sites on the @xmath0-symmetric structures .
we find that the axial ( reflection ) @xmath0 symmetry leads to reciprocal reflection ( transmission ) .
the transmission ( reflection ) is unidirectional affected by the magnetic flux and gain / loss , this is because the magnetic flux induced nonreciprocal phase and the gain / loss break the @xmath1 or @xmath2 symmetry of the scattering centre .
the results are further elucidated using a @xmath0-symmetric rhombic ring structure with enclosed effective magnetic flux describing by tight - binding model .
the physical realization of such scattering centre is possible in optical systems such as coupled waveguides array and coupled resonators .
notice that our conclusions are also applicable to the system with nonreciprocal tunnelling being unequal tunnelling amplitude @xcite .
we believe our findings may shed light on coherent light transport and would be useful for applications of quantum devices with inherent symmetry , in particular , for novel unidirectional optical device designs that not limited to optical diodes using synthetic @xmath0-symmetric metamaterial .
* schrdinger equations . *
the input and output leads are described by two semi - infinite tight - binding chain .
the left lead is @xmath152 , the right lead is @xmath153 , where @xmath154 ( @xmath155 ) is the creation ( annihilation ) operator of the site @xmath156 , the tunnelling between sites is uniform and set unity .
the hamiltonian of the scattering system is @xmath157 .
the eigenstate of the scattering system is set @xmath158 . for the axial @xmath0-symmetric configuration shown in fig .
[ fig1]c , the hamiltonian of the scattering system is @xmath159 .
the schrdinger equations @xmath160 on the scattering centre yield four independent equations@xmath161 for the reflection @xmath0-symmetric configuration shown in fig .
[ fig1]d , the hamiltonian of the scattering centre is @xmath162 .
correspondingly , four independent equations from the schrdinger equations @xmath163 on the scattering centre are in form of@xmath164where @xmath62 .
the schrdinger equations on the leads give the energy @xmath165 for the input with wave vector @xmath23 .
notice that @xmath166 in the reflection @xmath0-symmetric rhombic ring with @xmath167 ( @xmath56 ) results in @xmath168 and the transmissions are 1 .
otherwise , @xmath166 in system with @xmath169 ( @xmath56 ) leads to @xmath170 and the transmissions are 0 . *
reflection and transmission coefficients .
* to calculate the reflection and transmission coefficients , we set the left side input wave functions equation ( [ psi_l ] ) as @xmath171and the right side input wave functions equation ( [ psi_r ] ) as @xmath172 substituting @xmath173 , @xmath174 , @xmath175 , @xmath176 of equations ( [ fl]-[fr ] ) into equations ( [ l1]-[l4 ] ) and eliminating @xmath177 , @xmath178 , we get equations of @xmath21 , @xmath179 , @xmath25 , @xmath26 for the axial @xmath3-symmetric rhombic ring configuration . through directly algebraic calculation and simplification ,
we obtain the reflection and transmission coefficients @xmath21 , @xmath22 , @xmath25 , @xmath180 as functions of @xmath181 given in equations ( axialptrlr-[axialpttr ] ) . using the same procedure
, we get the reflection and transmission coefficients for the reflection @xmath0-symmetric rhombic ring configuration . after substituting @xmath173 , @xmath174 , @xmath182 , @xmath176 of equations ( [ fl]-[fr ] ) into equations ( [ r1]-[r4 ] ) and eliminating @xmath177 , @xmath178 , we get equations of @xmath21 , @xmath22 , @xmath25 , @xmath180 . through directly algebraic calculation and simplification ,
we obtain the reflection and transmission coefficients @xmath21 , @xmath183 , @xmath25 , @xmath26 as functions of @xmath184 given in equations ( [ reflectionptrl]-[reflectionpttlr ] ) . * scattering states at the spectral singularities .
* the scattering coefficients diverge at the spectral singularities , to calculate the scattering states , we have the wave functions of equation ( [ psi_l ] ) replaced by @xmath185 , @xmath186 . substituting @xmath173 , @xmath187 , @xmath175 , @xmath176 into equations ( [ l1]-[l4 ] ) of the axial @xmath0-symmetric rhombic ring configuration
, we obtain the coefficients satisfying @xmath188 at the spectral singularities that i ) @xmath109,@xmath189 ; and ii)@xmath190 , @xmath191 .
these indicate the scattering states are a self - sustained emission and a reflectionless absorption .
this work is partly supported by the national basic research program ( 973 program ) of china ( grant no .
2012cb921900 ) . l.j . also appreciates the support of nankai university baiqing plan foundation ( grant no .
zb15006104 ) .
x.z.z . also appreciates the support of national natural science foundation of china ( grant no .
11505126 ) and phd research startup foundation of tianjin normal university ( grant no .
52xb1415 ) .
z.s . also appreciates the support of cnsf ( grant no .
11374163 ) .
l.j . conceived the idea , carried out the study .
z.s . supervised the project . l.j . wrote the manuscript with helpful suggestions from x.z.z .
, g.z . , and z.s .
all authors discussed the results and reviewed the manuscript .
correspondence and requests for materials should be addressed to l. jin .
99 bender , c. m. & boettcher , s. real spectra in non - hermitian hamiltonians having @xmath0 symmetry .
lett . _ * 80 * , 5243 - 5246 ( 1998 ) .
bender , c. m. , brody , d. c. , & jones , h. f. complex extension of quantum mechanics .
lett . _ * 89 * , 270401 ( 2002 ) .
mostafazadeh , a. pseudo - hermiticity versus @xmath0 symmetry : the necessary condition for the reality of the spectrum of a non - hermitian hamiltonian . _ j. math .
phys . _ * 43 * , 205 - 214 ( 2002 ) .
znojil , m. spiked potentials and quantum toboggans .
_ j. phys . a : math .
* 39 * , 13325 - 13336 ( 2006 ) .
znojil , m. scattering theory with localized non - hermiticities .
d _ * 78 * , 025026 ( 2008 ) .
klaiman , s. , gnther , u. & moiseyev , n. visualization of branch points in @xmath0-symmetric waveguides .
lett . _ * 101 * , 080402 ( 2008 ) .
jin , l. & song , z. solutions of @xmath0-symmetric tight - binding chain and its equivalent hermitian counterpart .
rev . a _ * 80 * , 052107 ( 2009 )
. joglekar , y. n. , scott , d. , babbey , m. & saxena , a. robust and fragile @xmath0-symmetric phases in a tight - binding chain .
a _ * 82 * , 030103(r ) ( 2010 ) gong , j. & wang , q .- h .
geometric phase in @xmath0-symmetric quantum mechanics .
rev . a _ * 82 * , 012103 ( 2010 ) .
scott , d. d. & joglekar , y. n. @xmath0-symmetry breaking and ubiquitous maximal chirality in a @xmath0-symmetric ring .
rev . a _ * 85 * , 062105 ( 2012 )
. gong , j. & wang , q .- h .
time - dependent @xmath0-symmetric quantum mechanics .
_ j. phys . a : math .
theor . _ * 46 * , 485302 ( 2013 ) .
zhu , b. , l , r. & chen , s. @xmath0 symmetry in the non - hermitian su - schrieffer - heeger model with complex boundary potentials .
rev . a _ * 89 * , 062102 ( 2014 )
. wang , x. , liu , t. , xiong , y. & tong , p. spontaneous @xmath5-symmetry breaking in non - hermitian kitaev and extended kitaev models . _
rev . a _ * 92 * , 012116 ( 2015 )
. el - ganainy , r. , makris , k. g. , christodoulides , d. n. & musslimani , z. h. theory of coupled optical @xmath0-symmetric structures . _
lett . _ * 32 * , 2632 - 2634 ( 2007 ) .
makris , k. g. , el - ganainy , r. , christodoulides , d. n. & musslimani , z. h. beam dynamics in @xmath0 symmetric optical lattices .
lett . _ * 100 * , 103904 ( 2008 ) .
guo , a. _ et al .
_ observation of @xmath0-symmetry breaking in complex optical potentials .
* 103 * , 093902 ( 2009 ) .
rter , c. e. _ et al .
_ observation of parity - time symmetry in optics .
_ nature phys . _
* 6 * , 192 - 195 ( 2010 ) .
chong , y. d. , ge , l. , cao , h. & stone , a. d. coherent perfect absorbers : time - reversed lasers .
lett . _ * 105 * , 053901 ( 2010 ) .
longhi , s. @xmath0-symmetric laser absorber .
rev . a _ * 82 * , 031801(r ) ( 2010 ) . hasan , m. , ghatak , a. & mandal , b. p. critical coupling and coherent perfect absorption for ranges of energies due to a complex gain and loss symmetric system .
_ * 344 * , 17 - 28 ( 2014 ) .
regensburger , a. _ et al .
_ parity - time synthetic photonic lattices .
_ nature _ * 488 * , 167 - 171 ( 2012 ) .
jing , h. _ et al .
_ @xmath0-symmetric phonon laser .
lett . _ * 113 * , 053604 ( 2014 ) .
feng , l. , wong , z. j. , ma , r .-
m . , wang , y. & zhang , x. single - mode laser by parity - time symmetry breaking .
_ science _ * 346 * , 972 - 975 ( 2014 ) .
hodaei , h. , miri , m .- a . , heinrich , m. , christodoulides , d. n. & khajavikhan , m. parity - time - symmetric microring lasers
. _ science _ * 346 * , 975 - 978 ( 2014 ) .
peng , b. _ et al_. parity - time - symmetric whispering - gallery microcavities .
_ nature phys .
_ * 10 * , 394 - 398 ( 2014 ) .
peng , b. , zdemir , s. k. , chen , w. , nori , f. & yang , l. what is and what is not electromagnetically induced transparency in whispering - gallery microcavities .
commun . _ * 5 * , 5082 ( 2014 ) .
chang , l. _ et al_. parity - time symmetry and variable optical isolation in active - passive - coupled microresonators .
_ nature photon .
_ * 8 * , 524 - 529 ( 2014 ) .
jing , h. _ et al . _ optomechanically - induced transparency in parity - time - symmetric microresonators .
* 5 * , 9663 ( 2015 ) .
zhang , j. _ et al .
_ giant nonlinearity via breaking parity - time symmetry : a route to low - threshold phonon diodes .
_ phys . rev . b _ * 92 * , 115407 ( 2015 ) .
mostafazadeh , a. spectral singularities of complex scattering potentials and infinite reflection and transmission coefficients at real energies .
lett . _ * 102 * , 220402 ( 2009 ) .
longhi , s. spectral singularities and bragg scattering in complex crystals .
rev . a _ * 81 * , 022102 ( 2010 ) . mostafazadeh , a. optical spectral singularities as threshold resonances .
a _ * 83 * , 045801 ( 2011 ) . ghatak , a. , nathan , j. a. , mandal , b. p. & ahmed . z. spectral singularity and reflectionlessness in non - hermitian ( complex ) ginocchio potential .
_ j. phys . a : math .
theor . _ * 45 * , 465305 ( 2012 ) .
mostafazadeh , a. nonlinear spectral singularities for confined nonlinearities .
lett . _ * 110 * , 260402 ( 2013 ) .
mostafazadeh , a. nonlinear spectral singularities of a complex barrier potential and the lasing threshold condition .
rev . a _ * 87 * , 063838 ( 2013 ) . ramezani , h. , li , h .- k . ,
wang , y. & zhang , x. unidirectional spectral singularities . _
* 113 * , 263905 ( 2014 ) .
lin , z. _ et al_. unidirectional invisibility induced by @xmath0-symmetric periodic structures .
lett . _ * 106 * , 213901 ( 2011 ) .
feng , l. _ et al_. experimental demonstration of a unidirectional reflectionless parity - time metamaterial at optical frequencies . _
nature mater .
_ * 12 * , 108 - 113 ( 2013 ) .
longhi , s. invisibility in non - hermitian tight - binding lattices .
rev . a _ * 82 * , 032111 ( 2010 ) . longhi , s. a unidirectionally invisible @xmath5-symmetric complex crystal with arbitrary thickness .
_ j. phys . a : math .
theor . _ * 47 * , 485302 ( 2014 ) .
mostafazadeh , a. invisibility and @xmath0 symmetry .
rev . a _ * 87 * , 012103 ( 2013 ) . mostafazadeh , a. unidirectionally invisible potentials as local building blocks of all scattering potentials .
a _ * 90 * , 023833 ( 2014 ) .
mostafazadeh , a. active invisibility cloaks in one dimension .
a _ * 91 * , 063812 ( 2015 ) .
lin , z. , schindler , j. , ellis , f. m. & kottos , t. experimental observation of the dual behavior of @xmath0-symmetric scattering .
a _ * 85 * , 050101(r ) ( 2012 ) .
kalish , s. , lin , z. & kottos , t. light transport in random media with @xmath0 symmetry .
rev . a _ * 85 * , 055802 ( 2012 ) .
ahmed , z. reciprocity and unitarity in scattering from a non - hermitian complex @xmath0-symmetric potential .
a _ * 377 * , 957 - 959 ( 2013 ) .
mostafazadeh , a. generalized unitarity and reciprocity relations for @xmath0-symmetric scattering potentials .
_ j. phys . a : math .
theor . _ * 47 * , 505303 ( 2014 ) .
li , x. q. , zhang , x. z. , zhang , g. & song , z. asymmetric transmission through a flux - controlled non - hermitian scattering center .
rev . a _ * 91 * , 032101 ( 2015 ) . cannata , f. , dedonder , j .-
ventura , a. scattering in @xmath0-symmetric quantum mechanics . _
_ * 322 * , 397 - 433 ( 2007 ) .
longhi , s. aharonov - bohm photonic cages in waveguide and coupled resonator lattices by synthetic magnetic fields .
* 39 * , 5892 - 5895 ( 2014 ) .
fang , k. , yu , z. & fan , s. photonic aharonov - bohm effect based on dynamic modulation .
* 108 * , 153901 ( 2012 ) .
li , e. , eggleton , b. j. , fang , k. & fan , s. photonic aharonov - bohm effect in photon - phonon interactions .
* 5 * , 3225 ( 2013 ) .
fang , k. & fan , s. effective magnetic field for photons based on the magneto - optical effect .
rev . a _ * 88 * , 043847 ( 2013 ) . fang , k. , yu , z. & fan , s. realizing effective magnetic field for photons by controlling the phase of dynamic modulation .
_ nature photon . _
* 6 * , 782 - 787 ( 2012 ) .
chong , y. d. & liang , g. q. optical resonator analog of a two - dimensional topological insulator .
_ * 110 * , 203904 ( 2013 ) .
hafezi , m. synthetic gauge fields with photons .
b _ * 28 * , 1441002 ( 2014 ) .
jin , l. & song , z. hermitian scattering behavior for a non - hermitian scattering center .
a _ * 85 * , 012111 ( 2012 ) .
zhang , x. z. , jin , l. & song , z. self - sustained emission in semi - infinite non - hermitian systems at the exceptional point .
rev . a _ * 87 * , 042118 ( 2013 )
. zhang , g. , li , x. q. , zhang , x. z. & song , z. transmission phase lapse in the non - hermitian aharonov - bohm interferometer near the spectral singularity .
rev . a _ * 91 * , 012116 ( 2015 )
. zhang , x. z. & song , z. momentum - independent reflectionless transmission in the non - hermitian time - reversal symmetric system .
( n.y . ) _ * 339 * , 109 - 121 ( 2013 ) .
in the supplementary information , we show the reciprocal reflection ( transmission ) for axial ( reflection ) @xmath0 symmetry in details .
the input and output leads are described by two uniform semi - infinite tight - binding chains connected to the scattering centre .
the wave functions for the left and right lead input with wave vector @xmath23 are @xmath192 and @xmath17 , in form of @xmath193where @xmath194 and @xmath195 represent corresponding reflection and transmission coefficients of wave with vector @xmath23 .
in this situation ( fig .
[ fig1]a ) , the whole scattering system is axial @xmath3-symmetric with respect to the leads , where @xmath196 @xmath197 , @xmath198 @xmath199 , and @xmath200 @xmath201 .
the @xmath0 symmetry of the scattering system results in symmetric relations on its wave functions . in order to reveal the symmetry properties hidden in the scattering wave functions
, we act the @xmath0 operator on the wave function of left lead input @xmath202 , and get @xmath203we act @xmath0 operator on the wave function of right lead input @xmath204 , and get @xmath205the real energy scattering state is @xmath0-symmetric due to the @xmath5 symmetry of the whole scattering system . moreover , the two series of wave functions ( the eigenstates before and after acting the @xmath5 operator ) are both eigenstates of the system with eigenvalue @xmath206 .
therefore , they must be in accords with each other , i.e. , we can use @xmath202 and @xmath207 to compose @xmath208 and @xmath209 , because the left side input and right side input scattering states are degenerate .
composing @xmath210 via @xmath211 and @xmath212 of equations ( [ psi_l ] , [ psi_r ] ) by eliminating @xmath213 in @xmath214 region and comparing the coeffients of @xmath19 in the result with @xmath215 , we obtain@xmath216composing @xmath209 via @xmath211 and @xmath212 of equations ( [ psi_l ] , [ psi_r ] ) by eliminating @xmath217 in @xmath218 region and comparing the coeffients of @xmath19 in the result with @xmath219 , we obtain@xmath220we simplify the relations in equations ( [ relation_axial3 ] , relation_axial4 ) , and get@xmath221we simplify the relations in equations ( [ relation_axial1 ] , relation_axial2 ) , and get@xmath222from equations ( [ relation_axial5 ] , [ relation_axial7 ] ) , we obtain the reciprocal reflection for axial @xmath0 symmetry.@xmath32moreover , considering the wave vector @xmath23 and @xmath35 , we obtain @xmath223 , @xmath224 by comparing @xmath225 and @xmath226 ; and we obtain @xmath227 , @xmath39 by comparing @xmath228 and @xmath229 .
in this situation ( fig .
[ fig1]b ) , the whole scattering system is reflection @xmath5-symmetric with respect to the leads , where @xmath230 @xmath231 , @xmath232 @xmath233 , and @xmath234 @xmath201 . the @xmath0 symmetry of the scattering system results in symmetric relations on its wave functions .
in order to reveal the symmetry properties hidden in the scattering wave function , we act the @xmath0 operator on the wave function of the left lead input @xmath202 similarly as the previous analysis , and get@xmath235we act the @xmath0 operator on the wave function of right lead input @xmath236 , and get@xmath237we take @xmath202 and @xmath207 to compose @xmath208 and @xmath209 .
composing @xmath202 and @xmath207 of equations ( [ psi_l ] , [ psi_r ] ) in @xmath218 region by eliminating @xmath217 and comparing the coeffients of @xmath238 in the result with @xmath210 , we obtain@xmath239 composing @xmath202 and @xmath207 of equations ( [ psi_l ] , [ psi_r ] ) in @xmath214 region by eliminating @xmath213 and comparing the coeffients of @xmath19 in the result with @xmath209 , we obtain@xmath240we simplify the relations in equations ( [ relation_reflection1 ] , relation_reflection2 ) , and get@xmath241we simplify the relations in equations ( [ relation_reflection3 ] , relation_reflection4 ) , and get@xmath242from equations ( [ relation_reflection5 ] , [ relation_reflection7 ] ) , we obtain the reciprocal transmission for reflection @xmath0 symmetry.@xmath45moreover , considering the wave vector @xmath23 and @xmath35 , we obtain @xmath243 , @xmath244 by comparing @xmath245 and @xmath226 ; and we obtain @xmath246 , @xmath247 by comparing @xmath248 and @xmath229 . | parity - time ( @xmath0 ) symmetry is of great interest .
the reciprocal and unidirectional features are intriguing besides the @xmath0 symmetry phase transition . recently ,
the reciprocal transmission , unidirectional reflectionless and invisibility are intensively studied . here
, we show the reciprocal reflection / transmission in @xmath0-symmetric system is closely related to the type of @xmath0 symmetry , that is , the axial ( reflection ) @xmath0 symmetry leads to reciprocal reflection ( transmission ) .
the results are further elucidated by studying the scattering of rhombic ring form coupled resonators with enclosed synthetic magnetic flux .
the nonreciprocal phase shift induced by the magnetic flux and gain / loss break the parity ( @xmath1 ) and time - reversal ( @xmath2 ) symmetry but keep the parity - time ( @xmath0 ) symmetry .
the reciprocal reflection ( transmission ) and unidirectional transmission ( reflection ) are found in the axial ( reflection ) @xmath0-symmetric ring centre .
the explorations of symmetry and asymmetry from @xmath0 symmetry may shed light on novel one - way optical devices and application of @xmath0 metamaterials . |
the discovery of the integer quantum hall effect ( iqhe ) stimulates novel fundamental concepts in condensed matter physics , such as the gauge invariance,@xcite , edge state,@xcite and chern number@xcite .
in particular , y. hatsugai reveals the relationship between chern number and edge states in the iqhe,@xcite which provides another way from the edge state to understand the topological order in finite systems.@xcite moreover , the successful synthesis of nano and novel materials , such as the triangular organic material @xmath4,@xcite the kagome lattice herbertsmithite@xcite and the three - dimensional hyperkagome lattice magnet @xmath5,@xcite provide many opportunities to examine theoretically and experimentally some novel physical properties , including the topological properties,@xcite fractionalized excitation@xcite , singlet valence - bond solid states,@xcite , and edge states .
theoretically , these materials can be mapped to novel geometric lattices , such as the star lattice which is also called the triangle - honeycomb lattice@xcite , fisher lattice or decorated honeycomb lattice.@xcite these geometric lattice models provide a new view to understand the geometric effect and spin frustration.@xcite in particular , the spin models on the star and honeycomb lattices have shown many novel phases including the abelian and non - abelian anyons , chiral spin - liquid phases , topological orders , @xcite , magnetic orders@xcite , and topological insulator.@xcite the iqhe reveals some novel transport properties of electrons in two - dimensional ( 2d ) systems , in which the edge states play a key role and the hall conductance can be expressed in terms of chern number.@xcite interestingly , the bulk - edge correspondence discovered by y. hatsugai @xcite becomes an efficient method to explore the edge states in various 2d systems , such as the honeycomb lattice ( graphene)@xcite and spin - chiral ferromagnetic kagome lattice.@xcite however , the star lattice consists of a special lattice geometry , which can be mapped to a class of materials and cold atoms in optical lattices .
the mean field study of heisenberg model demonstrates the existence of several spin liquid phases , which depend on the flux configurations of two triangles and one dodecagon in the unit cell.@xcite a natural question arises : what is the relationship between the spin liquid phases and edge states on the star lattice with boundaries ? and @xmath6 .
there are six sites in the unit cell .
@xmath7 represents the hopping amplitude inside the triangle ( the bond in black ) , and @xmath8 corresponds to the hopping amplitude between different triangles ( the bond in blue).,width=288 ] in this paper , we focus on the edge states and their topological orders on the star lattice with boundaries .
we begin with a 2d tight - binding model with the hund s rule coupling .
it can be mapped to an effective tight - binding spinless model.@xcite the mean field approach predicts that there exists several spin liquid phases in the ground states with the local time reversal symmetry breaking.@xcite we use the bulk - edge correspondence method to analyze the edge states and their topological orders on the star lattice with boundaries .
this paper is organized as follows . in sec .
ii , we introduce the tight - binding model with boundaries and map it to a spinless tight - binding model . in sec .
iii , we give the bulk - edge correspondence for the star lattice .
we present the edge states and their corresponding chern numbers in various phases in sec .
finally , we give the discussion and conclusions . , @xmath9 , and one dodecagon .
the magnetic flux configurations are labeled by @xmath10 $ ] following ref .
in order to understand the relationship between the lattice geometry , edge states , and their topological properties , we consider the star lattice with boundaries , in which the conducting electrons move in a local spin background and couple with them by the hund s rule to form a double - exchange system . the corresponding tight - binding hamiltonian can be written as , @xmath11 where @xmath12 is the hopping amplitude between two nearest neighboring sites @xmath13 ; @xmath14 is the creation ( annihilation ) operator on site i with spin @xmath15 ; @xmath16 is the local spin on site i , which couples with the conducting electron spins with the effective coupling constant @xmath17 .
we consider that the local spins are approximately classical and the coupling @xmath17 is strong enough to have the hoping electrons to align them to the local spin @xmath18 on each site with the spinon function , @xmath19 , where @xmath20 are the spinon parameters . in this spinon representation , the hamiltonian eq.([eq : originalhamilton ] ) can be mapped to an effective tight - binding hamiltonian , @xmath21 where the effective hopping amplitude@xcite @xmath22e^{ia_{ij}}\nonumber\\ & = & t(\theta_{ij},\phi_{ij})e^{ia_{ij}}\end{aligned}\ ] ] where the phase @xmath23 is the vector potential generated by spin and corresponds to the berry phase felt by the hopping electron .
it is noted that the unit cell of the star lattice contains two triangular plaquettes and one 12-site dodecagon plaquette .
the mean field study of heisenberg model on the star lattice has revealed the existence of several spin liquid phases.@xcite the spin liquid phases depend on the flux figuration of the unit cell , which is labeled by the notation @xmath10$].@xcite in terms of the original spin variable , the fluxes on the triangular plaquettes correspond to the scalar spin chiralities @xmath24 , while @xmath25 is related to the 12 spins around the dodecagon loop . in the uniform spin liquid phase , @xmath26 $ ] , @xmath27 .
for the non - uniform spin liquid state , the spin chirality arises , the fermion hopping will acquire a phase @xmath28 , rendering a nonzero flux for a fermion moving around a loop @xmath29 .
@xmath30 , in principle , depends on the angles between spins @xmath16 and @xmath31 .
however in the mean field approximation , @xmath30 should be independent on the angles,@xmath32 for the spin liquid phases because the fluxes through plaquettes are periodic in the whole lattice.@xcite thus the hopping parameters @xmath33 can be classified into two independent variables on the star lattice .
@xmath7 labels the hopping amplitude inside the triangles and @xmath8 is the hopping amplitude between different triangles . for convenience ,
we introduce @xmath34 to measure the ratio of these two hopping amplitudes . for a given flux configuration , @xmath29
, there are different phase configurations @xmath23 .
the effect of the phase configuration can shift the whole energy band in the k space , but do not modify the energy band structure .
namely , different choices of the phase configurations do not change the edge state and their topological properties of the star lattice .
this allows us to set a simple choice of the phase configuration for various spin liquid phases to study their edge states and topological orders .
in general , we consider a strip of star lattice with the boundary along the @xmath35 direction and the periodic infinite @xmath36 direction shown in fig.1 .
we assume that the spin liquid phase @xmath10 $ ] , where the fluxes satisfy the constraint , @xmath37 . using the bloch theorem in the @xmath36 direction , @xmath38 , where @xmath39 labels the the unit cell and @xmath40 labels the sites in the unit cell .
we set @xmath41 for convenience .
the hamiltonian in eq.([effh ] ) can be written as @xmath42 where @xmath43 , and @xmath44_{n_{1}\times n_{1 } } \label{unitcell}\ ] ] with @xmath45 \label{unitcell } \\ % \end{equation } % \begin{equation } v&=&\left [ \begin{array}{cccccc } 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \\ r & 0 & 0 & 0 & 0 & 0% \end{array}% \right],\end{aligned}\ ] ] where @xmath46 is the number of unit cell along the @xmath35 direction . the bloch wave function can be written as @xmath47 , where @xmath39 runs the unit cell in the @xmath35 direction . inserting it into the schrdinger equation , @xmath48
, the solution can be reduced to a set of equations ( harper equation ) @xmath49 where @xmath50 . rewriting eq.([harper ] ) to a matrix form , we can express it in terms of a transfer matrix form , @xmath51 where @xmath52 is a @xmath53 matrix and its elements are @xmath54 where @xmath55 \\ l_2&=&1+r^4 - 2 ( 1+r^2 ) \varepsilon ^2+\varepsilon ^4 - 2 r^2 \cos k\\ l_3&=&\varepsilon(3 + 2 r^2+r^4 - 2(2+r^2 ) \varepsilon ^2+\varepsilon ^4)-2 r^2 \varepsilon\cos k\\ l_4&=&l_3 + 2 ( 1-\varepsilon ^2 ) \cos{\phi_1}-2 r^2 \cos(k+\phi_1)\\
l_5&=&l_3 + 2 ( 1-\varepsilon ^2 ) \cos\phi_2 - 2 r^2 \cos(k-\phi_2 ) \label{p}\end{aligned}\ ] ] we assume that the width @xmath56 of the star lattice contains an integer number of the unit cells , we can get the reduced transfer matrix @xmath57 considering the boundary condition @xmath58 , the edge energy @xmath59 satisfies @xmath60.@xcite for @xmath61 , the criterion for edge states follows @xcite @xmath62 the quantum hall conductance of systems can be expressed in terms of the chern number of u(1 ) bundle over the magnetic brillouin zone.@xcite the bulk - edge correspondence theory reveals that the chern number @xmath63 is equivalent to the winding number of the edge state moving around the hole of riemann surface ( the complex - energy surface ) , namely the intersection number between the canonical loop on the riemann surface and the trace of the edge state energy @xmath64.@xcite ( see fig .
thus , the quantum hall conductance can be given by the winding number of the edge states , @xmath65 when the fermi energy lies in the @xmath66th energy gap .
therefore , we can count the chern number from the energy spectrum of the system .
the mean field studies of the star lattice with the hamiltonian of eq.([eq : originalhamilton ] ) give various spin liquid phases.@xcite it is worth studying that the topological properties of these spin liquid phases .
$ ] for @xmath67 in ( a ) , @xmath68 in ( b ) , and @xmath69 in ( c).,width=336 ] the uniform spin liquid phase , @xmath26 $ ] respects all the space - group symmetry of the lattice and time reversal symmetry .
the energy spectra of the @xmath26 $ ] phase for several set of parameters @xmath70 are plotted in fig .
it can be seen that both of the bulk energy band and edge states are k - symmetric , @xmath71 , due to the space inverse symmetry , but the edge states are either embedded in the bulk states or isolated in the gap , namely there is no nontrivial bulk gap .
interestingly there exist two flat bands lying in the energy - band gap and touching an edge - state band , which is caused by interference .
@xcite it is similar to a uniform spin liquid on the kagome lattice and can be spoiled by perturbations , such as the next nearest neighbor hopping .
@xcite thus , the chern number is not well - defined and corresponds to common metals or insulators without iqhe .
actually the interactions from spinons can lead to instability of the uniform spin liquid phase to develop to the phases with breaking time reversal symmetry.@xcite the spin liquid phases are characterized by a set of spin chirality operators.@xcite different chiral spin phases exhibit different local magnetic fluxes .
for the nematic spin liquid phase @xmath72 $ ] , time reversal symmetry is broken spontaneously,@xcite we plot the energy spectra for some parameter @xmath70 in fig .
it can be seen that the k - symmetries of both bulk bands and edge states are broken , @xmath73 .
it shows that time reversal symmetry breaking could induce space inverse symmetry breaking that breaks the k - symmetries of bulk bands and edge states .
however , the edge states are also either embedded in the bulk states or isolated in the gap .
this implies the systems are common metals or insulators , but without iqhe , as @xmath26 $ ] phase .
$ ] for different parameters @xmath70.,width=336 ] for the chiral spin liquid phase i @xmath0 $ ] , time reversal symmetry is also broken spontaneously .
@xcite in principle , the chern number in the @xmath66th gap between the bulk energy bands depends on the parameters @xmath74 and @xmath1 , @xmath75 ( here we use this symbol for chern number ) .
however , we find from numerical investigations that the chern number in the range of @xmath74 and @xmath1 obeys the following symmetries : \(1 ) @xmath76 for @xmath77 ; \(2 ) @xmath78 for @xmath77 ; \(3 ) @xmath79 for @xmath80 ; thus , we can restrict the parameters only in the range of @xmath81 and @xmath82 . in fig .
6 we plot the energy spectrum for some typical parameters @xmath83 and @xmath84 .
the chern number can be counted by the winding number of the torus formed by two riemann surfaces.@xcite ( see fig .
the numbers in the right - hand side of each figures in figs .
6 are the chern number of the system when the fermi energy lies in the corresponding energy gap .
it can be seen that the bulk energy spectrum is k - symmetric , @xmath71 even though @xmath85 , but the edge state has no k - symmetry .
it implies that the space inversion symmetry still holds due to the bulk band k - symmetry , but the spontaneous time reversal breaking breaks the edge state k - symmetry .
$ ] for different parameters @xmath70.,width=336 ] different chern number implies different topological orders of the system . in order to give the phase diagram in the parameter space
, we try to find out the critical lines in the parameter space .
when the energy gaps close , the chern number must change , namely a phase transition happens .
notice that the bulk spectrum has the krammers degeneracy , @xmath71 , it implies that the close of the energy gap of the bulk band happens only at the @xmath86 point ( @xmath87 ) .
thus , we solve the eigenenergies of the unit cell hamiltonian ( bulk band ) at @xmath87 .
numerical investigation indicates the bulk energy bands at @xmath87 are linear with @xmath74 , which allows us suppose that the eigenenergies of the bulk energy band at @xmath87 have the form @xmath88 , where @xmath89 .
substituting this form into the eigen equation of the hamiltonian of the two - dimensional star lattice in the k space we can find that @xmath90 satisfies a cubic equation , @xmath91 when we consider @xmath81 , the solutions of the above cubic equation are @xmath92 which we set @xmath93 .
the closes of the bulk energy band gaps yield the critical lines in the parameter space , @xmath94 these three functions separate the parameter space @xmath95 into several different regions which have different topological orders shown in fig . 7 .
different regions in the phase diagram represent the ground states with different chern number configurations in which the chern number in different positions corresponds to the fermi energy in different gaps like fig .
the phases with different chern numbers reflect different integer quantum hall conductance of the system .
different region represents different chern numbers for various filling fraction , which are even functions of @xmath74 and periodic functions of @xmath1 with periodicity @xmath96 .
the energy spectrum along line 1 and line 2 are shown along ( a)@xmath97(d)@xmath97(b)@xmath97(c ) and ( d)@xmath97(e)@xmath97(f ) in fig .
6 respectively . ] for more general chiral spin liquid phases ii @xmath2 $ ] , the time reversal and space inverse symmetries are broken spontaneously .
for example , the energy spectra of the spin liquid phase @xmath98 $ ] for some specific parameters are shown in fig .
it can be seen that the k - symmetries of both bulk bands and edge state are broken @xmath73 , and the chern numbers in some energy gaps are not well - defined , such as the top three gaps for @xmath99 and @xmath100 .
because the degenerate points of the bulk band and edge state are not at the @xmath87 point , the phase transition lines can not be solved easily .
lcccccccc & eks & bks & trs & sis & & + & & & & & @xmath74 : & @xmath101 & @xmath102 & @xmath103 + @xmath26 $ ] & yes & yes & yes & yes & & @xmath104 & @xmath104 & x + @xmath105 $ ] & no & no & no & no & & x & x & @xmath106 + @xmath107 $ ] & no & yes & no & yes & & @xmath104 & @xmath104 & @xmath106 + @xmath108 $ ] & no & no & no & no & & @xmath106 & @xmath104 & @xmath106 + + +
to compare the basic energy - band properties of different spin liquid phases , we assume that the fermi energy lies in the middle of the middle energy - band gap . the energy band symmetry and chern numbers for some specific cases are listed in table i. it can be seen from table i that for @xmath26 $ ] both of the edge k - symmetry ( eks ) and bulk k - symmetry ( bks ) are held due to the time reversal invariance and space inversion invariance . when @xmath109 , the chern number becomes non - well - defined .
the cases in the @xmath110 lines in table i indicate that there is no eks , but bks remains for @xmath107 $ ] .
this implies that the spontaneous time reversal symmetry breaking does not always induce the space inversion symmetry breaking and the eks can be broken by the time reversal symmetry .
the results in table i reveal that the chern number depends on not only the time reversal and space inverse symmetries , but also the parameters @xmath111 of the star lattice .
the ground states become normal metals or semiconductors for the phases without the well - defined chern number .
interestingly , there is a topological invariance for the exchange of the magnetic fluxes in the two triangles @xmath2 $ ] and @xmath3 $ ] .
they have the same chern numbers and their bulk energy bands are k - asymmetric .
these findings indicate some new phases in the 2-dimensional materials.@xcite $ ] for different parameters @xmath70.,title="fig:",width=316][t ]
in summary , we have studied the edge states and their topological orders in the different spin liquid phases of star lattice by using the bulk - edge correspondence theory .
the bulk and edge - state energy structures and chern number depend on the spin liquid phases and hopping parameters because the local spontaneous magnetic flux in the spin liquid phases breaks the time reversal and space inversion symmetries .
we have given the characteristics of bulk and edge energy structures and their corresponding chern numbers in the uniform , nematic and chiral spin liquid phases .
in particular , we have obtained analytically the phase transition lines of different topological phases and their corresponding phase diagrams for the chiral spin liquid states @xmath0 $ ] .
we have also found that the topological invariance for the spin liquid phases,@xmath2 $ ] and @xmath3 $ ] .
the results tell us the relationship between the energy - band and edge - state structures and their topological orders of the star lattice .
especially , this star lattice has been synthesized in the material called iron acetate recently.@xcite therefore , our results provide a hall conductance experimental guideline to discriminate the spin liquid phases in real materials and cold atoms in optical lattice .
the changes of filling fraction could be implemented by tuning the applied gate voltage .
these results can also give some hints for understanding the heisenberg model on the star lattice .
we thank ming - liang tong and xiao - ming chen for helpful discussions .
y . huang thanks an zhao and jie - sen li for useful discussions on numerical calculations .
this work is supported by the fundamental research funds for the central universities of china ( 11lgjc12 and 10lgzd09 ) , nsfc-11074310 , most of china 973 program ( 2012cb821400 ) , specialized research fund for the doctoral program of higher education ( 20110171110026 ) , and ncet-11 - 0547 .
r. b. laughlin , phys .
b 23 , 5632(1981 ) b. i. halperin , phys .
b 25 , 2185 ( 1982 ) d. j. thouless , m. kohmoto , m. p. nightingale and m. den nijs , phys . rev .
49 , 405 ( 1982 ) y. hatsugai , phys
71 , 3697 ( 1993 ) .
y. hatsugai , phys.rev .
b 48 , 11851 ( 1993 ) . zhigang wang , ping zhang , new j. phys .
11 , 123014 ( 2010 ) y. shimizu , k. miyagawa , k. kanoda , m. maesato and g. saito , phys .
91 , 107001 ( 2003 ) matthew p. shores , emily a. nytko , bart m. bartlett and daniel g. nocera , j. am .
127 , 13462(2005 ) m. sasaki , k. hukushima , h. yoshino and h. takayama , phys .
99 , 137202 ( 2007 ) yi zhou , patrick a. lee , tai - kai ng and fu - chun zhang , phys .
lett . 101 , 197201 ( 2008 ) michael j. lawler , arun paramekanti , yong baek kim and leon balents , phys .
lett . 101 , 197202 ( 2008 ) p. nikolic and t. senthil , phys . rev .
b 68 , 214415 ( 2003 ) rajiv r. p. singh and david a. huse , phys .
b 76 , 180407(r ) ( 2007 ) j. b. marston and c. zeng , j. appl .
phys . 69 , 5962 ( 1991 ) a. kitaev , ann .
321 , 2 ( 2006 ) .
h. yao and s. a. kivelson , phys .
99 , 247203 ( 2007 ) .
g. kells , d. mehta , j. k. slingerland , and j. vala , phys . rev .
b 81 , 104429 ( 2010 ) . k. ohgushi , s. murakami , and n. nagaosa , phys .
b 62 , r6065 ( 2000 ) b .- j .
yang , a. paramekanti , y.b .
kim , phys .
b 81 , 134418(2010 ) .
t. p. choy and y. b. kim , phys .
b 80 , 064404 ( 2009 ) .
a. regg , j. wen , and g. a. fiete , phys . rev .
b 81 , 205115 ( 2010 ) .
jun wen , andreas regg , joseph c .- c .
wang and gregory a. fiete phys .
b 82 , 075125 ( 2010 ) y. hatsugai , t. fukui , and h. aoki , phys .
b 74 , 205414 ( 2006 ) zhigang wang and ping zhang phys .
b 77 , 125119 ( 2008 ) | a group of novel materials can be mapped to the star lattice , which exhibits some novel physical properties .
we give the bulk - edge correspondence theory of the star lattice and study the edge states and their topological orders in different spin liquid phases .
the bulk and edge - state energy structures and chern number depend on the spin liquid phases and hopping parameters because the local spontaneous magnetic flux in the spin liquid phase breaks the time reversal and space inversion symmetries .
we give the characteristics of bulk and edge energy structures and their corresponding chern numbers in the uniform , nematic and chiral spin liquids .
in particular , we obtain analytically the phase diagram of the topological orders for the chiral spin liquid states @xmath0 $ ] , where @xmath1 is the magnetic flux in two triangles and a dodecagon in the unit cell .
moreover , we find the topological invariance for the spin liquid phases , @xmath2 $ ] and @xmath3 $ ] .
the results reveal the relationship between the energy - band and edge - state structures and their topological orders of the star lattice . |
let @xmath0 be a unital complex banach algebra . a linear mapping @xmath1 from a subspace @xmath2 into another unital complex banach algebra @xmath3
is called _ spectrally bounded _ if there is a constant @xmath4 such that @xmath5 for all @xmath6 . here , and in what follows , @xmath7 stands for the spectral radius of a banach algebra element @xmath8 .
this concept evolved in banach algebra theory , and especially automatic continuity , over time in the 1970 s and 1980 s but the terminology was only introduced in @xcite , together with its companions _ spectrally infinitesimal _ : @xmath9 ; _ spectrally contractive _ : @xmath10 ; and _ spectrally isometric _ : @xmath11 for all @xmath8 .
it follows from @xcite , see also @xcite , lemma a , that the separating space of every surjective / @xmath12 on a closed subspace @xmath13 is contained in the radical of @xmath3 ; thus @xmath12 is bounded if @xmath3 is semisimple .
this was used by aupetit in @xcite to give a new proof of johnson s uniqueness - of - the - complete - norm - topology theorem ; cf .
also @xcite .
it was further exploited in @xcite to investigate continuity properties of lie epimorphisms .
a systematic study of /s was begun in @xcite , with its main results published in @xcite and @xcite . since then , the interest in the topic has steadily grown and by now there is a sizeable literature on a variety of aspects . the present paper s aim is to discuss several of the open problems on /s along with some new results that should make it clear why these questions are natural and important .
section [ sec : auto - cont ] is devoted to a recapitulation of probably the most important and deepest problem on /s , the non - commutative singer wermer conjecture .
no substantial progress seems to have been made on this over the last few years .
a standard method to reduce a more sophisticated problem in banach algebra theory to a simpler one is to use quotients .
when dealing with operators between banach algebras , we , of course , need invariant ideals to perform this . in section [ sec : quotients ] we discuss the interplay between properties of a ( / ) operator and the operator it induces on suitable quotient banach algebras in a fairly general setting and recover recent results for operators preserving the essential spectral radius on @xmath14 in @xcite .
the identity element plays a distinguished role for /s .
suppose that @xmath15 is a surjective / between unital s @xmath0 and @xmath3 .
since @xmath12 restricted to the centre @xmath16 of @xmath0 is a * -isomorphism onto the centre @xmath17 of @xmath3 ( @xcite , proposition 2.3 ) , @xmath18 is a central unitary in @xmath3 .
therefore , whenever we study a / @xmath12 , we can without loss of generality assume that @xmath12 is unital ( i.e. , @xmath19 ) ; see the proof of @xcite , corollary 2.6 . for a general
/ , the situation changes and for this reason we shall pay attention in section [ sec : value - at - one ] to the relevance of the value @xmath18 .
as it turns out , if the domain algebra is sufficiently ` infinite ' , then @xmath18 must be a central invertible element in @xmath3 . as a consequence ,
we are able to extend the main results in @xcite and @xcite to the non - unital setting ( theorem [ thm : nonunital - spbdd ] below ) .
the special case @xmath20 , @xmath21 for hilbert spaces @xmath22 and @xmath23 was studied previously in @xcite . in @xcite
, we propose a non - selfadjoint version of kadison s classical theorem stating that every unital surjective isometry between unital s is a jordan * -isomorphism . in many cases , this conjecture ( see problem [ prob : spiom - conjecture ] below ) has been confirmed , however almost always under some strong assumption of ` infiniteness ' on the domain . in section [ sec : spec - isoms ] , we employ a reduction method developed in @xcite and @xcite to obtain a new result of a similar ilk ( theorem [ thm : no - tracial - states ] ) .
more importantly , perhaps , we propose a new route based on a description of /s in the presence of a trace to solve this open question at least in the setting of 1f / s ( corollary [ cor : type - twoone ] ) .
the _ separating space _ @xmath24 of an operator @xmath12 between normed spaces @xmath13 and @xmath25 is defined by @xmath26 and measures the degree of discontinuity of @xmath12 .
if @xmath13 and @xmath25 are banach spaces then @xmath12 is bounded if and only if @xmath27 , by the closed graph theorem . for a detailed discussion of the separating space see @xcite .
it follows from @xcite , see also @xcite and @xcite , that @xmath28 consists of quasinilpotent elements whenever @xmath1 is a / on a subspace @xmath13 of a banach algebra . in fact ,
@xmath29 for each @xmath6 and @xmath30 .
this , together with zemnek s characterisation of the jacobson radical @xmath31 of @xmath0 , entails the following result .
[ thm : aupetit - aut - cont ] let @xmath1 be a / on a closed subspace @xmath13 of a banach algebra onto a banach algebra @xmath3 . then @xmath32 . in particular ,
if @xmath3 is semisimple , then @xmath12 is bounded .
the above estimate on the spectral radii is used in @xcite to describe the continuity of surjective lie homomorphisms as follows .
[ thm : aut - cont - lie ] the separating space @xmath33 of a lie epimorphism @xmath34 between two banach algebras @xmath0 and @xmath3 is contained in @xmath35 , the centre modulo the radical .
the centre modulo the radical , @xmath35 , is the pre - image of the centre of @xmath36 under the canonical map @xmath37 and turns out to be the largest quasinilpotent lie ideal of @xmath3 . as a consequence , it is invariant under every lie epimorphism and thus the above result yields the following automatic continuity statement ( @xcite , corollary ) .
let @xmath38 be a lie epimorphism between the banach algebras @xmath0 and @xmath3 .
the induced lie epimorphism @xmath39 between the banach lie algebras @xmath40 and @xmath41 is continuous .
the centre modulo the radical also plays an important role in the automatic continuity of derivations on banach algebras .
it is an open question whether the separating space of each of the iterates @xmath42 , @xmath43 of a derivation @xmath44 on a banach algebra @xmath0 is contained in the radical of @xmath0 .
if this is the case , the following long - standing problem would have a positive answer .
[ prob : ncsw ] does @xmath45\in\mathcal{z}(a)$ ] for all @xmath46 imply that @xmath47 ? here ,
@xmath48 $ ] of course stands for the commutator @xmath49 .
if @xmath50 is replaced by the proper centre @xmath16 of @xmath0 , then problem [ prob : ncsw ] has an affirmative answer , due to a reduction technique developed in @xcite and marc thomas famous theorem @xcite .
the connection to /s is provided by the next result @xcite .
a derivation on a banach algebra @xmath0 is / if and only if it maps into @xmath31 . hence , problem [ prob : ncsw ] is equivalent to the following question .
* problem 1@xmath51 .
* does @xmath45\in\mathcal{z}(a)$ ] for all @xmath46 imply that @xmath52 is / ?
it appears that no progress on this question has been made in the last decade . for a fuller discussion
see @xcite and @xcite .
in the previous section , we exploited the concept of /s to study linear mappings satisfying some additional algebraic conditions , such as lie homomorphisms , derivations , etc .
another main direction of research has been on the question which algebraic properties can be derived from the assumption of spectral boundedness .
maybe the most prominent of these problems is the following one from @xcite . by a _
jordan epimorphism _ we will , of course , understand a surjective linear mapping preserving the jordan product @xmath53 .
[ prob : kaplansky ] let @xmath0 and @xmath3 be semisimple unital banach algebras .
suppose @xmath15 is unital , surjective and invertibility - preserving .
is @xmath12 necessarily a jordan epimorphism ?
note that a jordan epimorphism necessarily has all the properties assumed in problem [ prob : kaplansky ] and hence is bounded , by theorem [ thm : aupetit - aut - cont ] .
many contributions on kaplansky s problem have been made over the past decades but so far it eludes a final answer . from aupetit s substantial work on the question we only quote the following beautiful result in @xcite ; see also the references therein .
[ thm : aupetit - main ] let @xmath0 and @xmath3 be von neumann algebras .
then every unital surjective invertibility - preserving linear mapping @xmath15 is a jordan epimorphism . to the best of our knowledge ,
problem [ prob : kaplansky ] is still open even in the case when both domain and codomain are s ; cf .
also @xcite . the structural investigation of /s is partly motivated by the question to what extent the hypotheses in kaplansky s problem can be relaxed while retaining the same goal , to show that the mapping is a jordan homomorphism .
the relevance of the value at @xmath54 will be discussed in the next section . when considering the possibility of replacing `` invertibility - preserving '' by `` / '' we must keep in mind that ( a ) _ every _ bounded linear mapping defined on a commutative is / ( since spectral radius and norm coincide in spaces of the form @xmath55 ) ; ( b ) a finite trace on a is a spectral contraction , hence , already on the @xmath56 matrices there is a unital / onto @xmath57 which is not a jordan epimorphism . notwithstanding this
there are satisfactory results in the setting of ` very infinite ' s ; see sections [ sec : value - at - one ] and [ sec : spec - isoms ] .
many examples are known illustrating the fact that no strong results can be expected for non - surjective /s in general ; see , e.g. , @xcite .
we shall now study the situation when a / is merely surjective ` up to ' or ` modulo ' an ideal .
suppose @xmath58 and @xmath59 are proper closed ideals in the unital banach algebras @xmath0 and @xmath3 , respectively .
we say that a linear mapping @xmath15 is _ essentially / _
( more precisely , @xmath60-@xmath61-_essentially / _ ) if there is a constant @xmath4 such that @xmath62 for all @xmath46 .
we call @xmath12 _ surjective modulo _ @xmath61 if , for every @xmath63 , there is @xmath46 such such @xmath64 .
if @xmath65 , we can define the induced linear mapping @xmath66 by @xmath67 , @xmath46 .
the following proposition relates the properties of @xmath68 to those of @xmath12 .
[ prop : ess - spec - bdd ] let @xmath0 and @xmath3 be unital banach algebras .
suppose that @xmath60 and @xmath61 are proper closed ideals of @xmath0 and @xmath3 , respectively , such that @xmath69 is semisimple .
for a linear mapping @xmath15 the following conditions are equivalent . 1 .
@xmath12 is essentially spectrally bounded and surjective modulo @xmath61 ; 2 .
@xmath65 and @xmath68 is / and surjective .
( b)@xmath70(a ) under the assumption @xmath65 , @xmath68 is a well - defined linear mapping which is / if and only if @xmath12 is essentially / , by definition .
moreover , @xmath68 surjective precisely means that , for each @xmath63 , @xmath71 for some @xmath46 ; that is , @xmath12 is surjective modulo @xmath61 .
( a)@xmath70(b ) by the first paragraph in this proof , it suffices to show that @xmath65 .
take @xmath72 . by hypothesis , for each @xmath63 , there is @xmath73 such that @xmath74 .
let @xmath75 .
we have @xmath76 for some @xmath4 .
consequently , the subharmonic function @xmath77 is bounded on @xmath57 , hence constant .
it follows that @xmath78 for all @xmath63 from which we conclude that @xmath79 , by zemnek s characterisation of the radical .
as @xmath69 is semisimple , we obtain that @xmath80 as desired .
as an immediate consequence we have the following variant of the main result in @xcite .
[ cor : pure - infinite - quotient ] let @xmath15 be a unital linear mapping from a unital purely infinite @xmath0 with real rank zero into a unital banach algebra @xmath3 .
let @xmath59 be a proper closed ideal in @xmath3 such that @xmath69 is semisimple and suppose that , for some proper closed ideal @xmath58 of @xmath0 , the operator @xmath12 is @xmath60-@xmath61-essentially / and surjective modulo @xmath61 . then @xmath12 is a jordan epimorphism modulo @xmath61 .
by proposition [ prop : ess - spec - bdd ] , @xmath65 and the induced mapping @xmath66 is unital , / and surjective .
since @xmath81 is purely infinite and has real rank zero , corollary 2.5 in @xcite implies that @xmath68 is a jordan epimorphism ; hence , @xmath12 is a jordan epimorphism modulo @xmath61 .
in particular , if @xmath12 is even a spectral isometry modulo @xmath61 in the above situation , then @xmath68 provides a jordan isomorphism between the quotients @xmath81 and @xmath69 .
this was obtained in the special case @xmath82 and @xmath83 for an infinite dimensional hilbert space @xmath22 in @xcite , theorem 3.1 . in order to make use of quotients ,
invariant ideals are needed . in the setting of s , so - called glimm ideals offer themselves as good candidates , see section [ sec : spec - isoms ] below .
however , so far their invariance has only been established under additional hypotheses , for instance for / on von neumann algebras @xcite .
does every unital / from a unital onto another one leave each glimm ideal invariant ? the setting of s is favourable also because a / induces a / on every quotient ( @xcite , proposition 9 ) . in a more general
setting it is easy to see that a / induces a / quotient operator provided the domain algebra is an _
sr_-algebra ; cf . @xcite .
a surjective jordan homomorphism between unital algebras attains the value @xmath54 at @xmath54 . in this section
, we shall discuss the behaviour of an arbitrary / at @xmath54 to see how this affects the possible difference from a jordan homomorphism .
remembering that _ every _ bounded operator from a space @xmath55 is / , we have to be careful not to expect too much in a general setting .
the following observation follows directly from our earlier results .
[ prop : basic - fact ] let @xmath15 be a / from a unital @xmath0 onto a unital semisimple banach algebra @xmath3 . for every pair @xmath84 of mutually orthogonal properly infinite projections in @xmath0
we have @xmath85 by @xcite , corollary 3.2 , @xmath12 preserves elements of square zero .
by @xcite , proposition 2.1 , every element in the corners @xmath86 and @xmath87 can be written as a finite sum of elements of square zero ( in @xmath86 and @xmath87 , respectively ) , since both @xmath88 and @xmath89 are properly infinite .
the claim thus follows from lemma 3.3 in @xcite .
[ cor : t - of - one ] let @xmath15 be a / from a unital @xmath0 with real rank zero onto a unital semisimple banach algebra @xmath3 .
suppose that every non - zero projection in @xmath0 is properly infinite .
then @xmath18 is an invertible element in the centre of @xmath3 .
the basic idea of the argument has been used before in some special cases , see , e.g. , @xcite .
let @xmath90 be a non - trivial projection . applying the identity in proposition [ prop : basic - fact ] to @xmath91 and @xmath92 we obtain @xmath93 that is , @xmath94 . upon multiplying this identity first on the left , then on the right by @xmath95 and subtracting the resulting two identities we obtain @xmath96 for every ( non - trivial ) projection @xmath88 .
let @xmath97 be an orthogonal family of projections in @xmath0 . by applying ( [ eq : basic - fact ] ) inductively we find that @xmath98 . since @xmath0 has real rank zero , it follows that @xmath99 for all @xmath100 in a dense subset of @xmath101 and since @xmath12 is bounded , thus for all @xmath102 . as @xmath103 and @xmath104 for all selfadjoint @xmath105 , we conclude that @xmath18 commutes with @xmath106 for every @xmath46 .
the surjectivity of @xmath12 entails that @xmath18 commutes with every square of an element in @xmath3 but since @xmath107 for each @xmath63 , we finally obtain that @xmath18 belongs to the centre of @xmath3 .
going back to the first paragraph of the proof we therefore have @xmath108 for every projection @xmath90 . using the same argumentation as above and the fact that @xmath0 has real rank zero another time
we conclude that @xmath109 for all @xmath46 .
as @xmath12 is surjective , it follows that @xmath18 must be invertible .
whenever @xmath12 is a / such that @xmath18 is an invertible element in the centre of the codomain , the mapping @xmath110 defined by @xmath111 , @xmath46 is a unital / ; thus we can apply the results known in this situation .
[ thm : nonunital - spbdd ] let @xmath15 be a / from a unital @xmath0 onto a unital semisimple banach algebra @xmath3 .
suppose that 1 .
@xmath0 is a properly infinite von neumann algebra , or 2 .
@xmath0 is a purely infinite with real rank zero .
then there is a unique jordan epimorphism @xmath112 such that @xmath113 for all @xmath46 .
moreover , @xmath18 is a central invertible element in @xmath3 .
case ( ii ) is immediate from corollary [ cor : t - of - one ] , since every non - zero projection in a purely infinite is properly infinite .
thus we can define @xmath61 by @xmath114 , @xmath46 , which is a jordan epimorphism by @xcite , corollary 2.5 .
the case ( i ) will not only need the result in the unital case , @xcite , theorem 3.6 , in the same manner as just explained but also an elaboration of the projection techniques used in the main technical result to obtain the unital case , which is @xcite , proposition 3.4 .
note that , by the proof of corollary [ cor : t - of - one ] , it suffices to show that @xmath18 commutes with @xmath115 for every projection @xmath90 ; once this is verified , @xmath18 will be a central invertible element .
we shall use the same strategy as in proposition 3.4 of @xcite .
suppose first that both @xmath88 and @xmath116 are properly infinite .
then the assertion follows immediately from identity ( [ eq : basic - fact ] ) .
next suppose that @xmath88 is properly infinite but @xmath116 is not . by the halving lemma , there is a subprojection @xmath117 of @xmath88 such that @xmath118 , where @xmath119 denotes murray von neumann equivalence .
hence , all projections @xmath117 , @xmath120 , @xmath121 and @xmath122 are properly infinite , see the proof of @xcite , proposition 3.4 . by our first step
we thus have @xmath123 applying ( [ eq : basic - fact ] ) to @xmath117 and @xmath121 we have @xmath124 , wherefore @xmath18 commutes with @xmath115 .
suppose now that @xmath88 is infinite but not properly infinite .
then there is a non - trivial central projection @xmath125 in @xmath0 such that @xmath126 is properly infinite and @xmath127 is finite .
we need the following preliminary observation .
suppose that @xmath89 is a properly infinite projection in @xmath0 but @xmath128 is not .
choosing the subprojection @xmath117 of @xmath89 as in the last paragraph we have @xmath129 where the last two equality signs come from identity ( [ eq : basic - fact ] ) . multiplying this identity first on the left , then on the right by @xmath18 and subtracting the two identities we obtain @xmath130 by the previous paragraph .
to simplify our calculations we will now use the usual commutator notation @xmath48=xy - yx$ ] . since @xmath126 is a properly infinite projection , the preliminary observation yields @xmath131=0 $ ] . by the proof of @xcite , proposition 3.4 , the projection @xmath132 is properly infinite as well ; therefore , @xmath133=0 $ ] , too . as a result , @xmath134 & = [ ( t1)^2,tp]-[(t1)^2,t(zp ) ] = -\ , [ ( t1)^2,t(1-p)]\\ & = -\ , [ ( t1)^2,t(z(1-p))]-[(t1)^2,t((1-z)(1-p))]\\ & = -\ , [ ( t1)^2,tz]+[(t1)^2,t(zp)]=0 , \end{split}\ ] ] since every non - zero central projection in @xmath0 , in particular @xmath125 , is properly infinite and thus @xmath135=0 $ ] as well . from identity ( [ eq : basic - fact ] ) we have @xmath136 since @xmath137 is non - zero , hence properly infinite
. in particular , @xmath138 it follows that @xmath139 combining this with @xmath140=t1\,[t1,tw]+[t1,tw]\,t1=[t1,t1\,tw+tw\,t1]\qquad(w\in a)\ ] ] we find that @xmath141\\ & = [ t1,t(1-z)^2]+[t1,t((1-z)p)^2 ] - [ t1,(t1)\,t((1-z)p)+t((1-z)p)\,(t1)]\\ & = 0 + [ t1,t((1-z)p)^2 ] - [ ( t1)^2,t((1-z)p ) ] , \end{split}\ ] ] that is , @xmath142=[(t1)^2,t((1-z)p]$ ] .
the commutator on the right hand side is zero , as we saw above .
therefore , @xmath143=[t1,t(zp)^2]+[t1,t((1-z)p)^2]=0,\ ] ] where we used ( [ eq : basic - fact - central ] ) once again .
finally suppose that @xmath88 is finite .
then @xmath144 is infinite .
letting @xmath145 be the central projection such that @xmath146 is properly infinite and @xmath147 is finite , we recollect the necessary information from the arguments above . 1 . @xmath148=0 $ ] ; 2 .
@xmath149=[t1 , t((1-z')p')^2]=0 $ ] ; 3 .
@xmath150=[t1,t(z'p')^2]=0 $ ] .
consequently , @xmath151=[(t1)^2,t(z'p')]+[(t1)^2,t((1-z')p')]=0.\ ] ] as @xmath152 & = [ t1 , ( t1)^2]+[t1,(tp)^2]-[t1,(t1)\,(tp)+(tp)\,(t1)]\\ & = [ t1,(tp)^2]-[(t1)^2,tp]=[t1,(tp)^2]+[(t1)^2,t(1-p ) ] , \end{split}\ ] ] we conclude that @xmath153=[t1,(tp')^2]-[(t1)^2,tp']=0.\ ] ] this completes the argument that @xmath154=0 $ ] for every projection @xmath90 and , as explained above , this is sufficient to prove the result . evidently , all we need to assume on @xmath12 in the proof of the above theorem is that @xmath12 is surjective , bounded and preserves elements of square zero .
thus , the main result in @xcite , theorem 2.4 , extends appropriately to the non - unital setting . as pointed out above
, we can not expect a result like theorem [ thm : nonunital - spbdd ] for arbitrary s of real rank zero or even von neumann algebras .
however , in a more restricted setting the situation might change .
[ prob : tone - on - finite - factor ] let @xmath12 be a / defined on a finite von neumann factor ( onto a semisimple unital banach algebra @xmath3 )
. is @xmath18 a non - zero complex multiple of the identity in @xmath3 ?
generalising the classical banach stone theorem , kadison showed in @xcite that every unital surjective isometry @xmath12 between two unital s must be a jordan * -isomorphism .
in fact , he showed a stronger result in @xcite , theorem 2 , namely that it suffices that @xmath12 maps the selfadjoint part @xmath101 of @xmath0 isometrically onto the selfadjoint part @xmath155 of @xmath3 .
there are many interesting consequences of these results , all relating isometric properties to selfadjointness . as an example , we state and prove a variant of a theorem of chan ( @xcite , theorem 3 ) . since @xmath158 is a norm , @xmath12 is injective and it is clear that thus @xmath159 is numerical radius - preserving as well . as @xmath160 for all @xmath161 , by kadison
s theorem it suffices to show that @xmath162 .
let @xmath102 , @xmath163 .
write @xmath164 with @xmath165 .
suppose @xmath166 ; then there is @xmath167 and we may assume that @xmath168 . for each state
@xmath169 on @xmath3 and each @xmath43 , we have @xmath170 hence , for large @xmath171 , @xmath172 which is impossible as @xmath12 is numerical radius - preserving . therefore , @xmath173 and @xmath174 is selfadjoint .
kadison s theorem in one direction and an application of the russo
dye theorem in the other direction show that a unital surjective spectral isometry is selfadjoint ( i.e. , maps selfadjoint elements to selfadjoint elements ) if and only if it is an isometry , see @xcite , proposition 2.4 . these results and others , and the analogy to kaplansky
s question ( problem [ prob : kaplansky ] above ) , made us ask the following question which we indeed state as a conjecture in @xcite .
[ thm : no - tracial - states ] let @xmath15 be a unital spectral isometry from a unital @xmath0 with real rank zero and without tracial states such that @xmath175 is hausdorff and totally disconnected onto a unital @xmath3 .
then @xmath12 is a jordan isomorphism .
there are ( at least ) five structure spaces associated with a unital @xmath0 , which we will briefly discuss .
their relation can be depicted as follows .
@xmath176 the set @xmath177 of all irreducible representations of @xmath0 maps onto the set @xmath178 of all unitary equivalence classes of such representations .
since equivalent representations have the same kernel , there is a canonical surjection from @xmath178 onto @xmath175 , the set of all primitive ideals of @xmath0 .
the latter carries a natural topology , the jacobson topology , which can be pulled back to @xmath178 , and then to @xmath177 , to turn them into topological spaces in a canonical way . since @xmath16 is a commutative unital , @xmath179 allows several equivalent descriptions of which we choose the one via maximal ideals . for each maximal ideal @xmath180 of @xmath16 , the closed ideal @xmath181
is called a _ glimm ideal _ of @xmath0 . for each @xmath182
, @xmath183 gives the generating maximal ideal in @xmath16 back , wherefore there is a bijection between the two sets .
transporting the natural topology of @xmath179 over to @xmath184 thus turns the latter into a compact hausdorff space . from the other direction
, we can apply the _ complete regularisation map _
@xmath185 defined by @xmath186 and @xmath187 is the equivalence class with respect to this relation .
if @xmath60 is the glimm ideal given by @xmath188 then @xmath187 can be identified with @xmath189 . among the consequences of this
is @xmath190 , that is , the glimm ideals separate the points of @xmath0 .
it follows from the definition of the jacobson s topology that @xmath175 is t@xmath191 if and only if every primitive ideal is maximal . in a similar vein , @xmath175 is t@xmath192 , i.e. , hausdorff , if and only if every glimm ideal is maximal ( @xcite , lemma 9 ) .
in contrast to results like theorem [ thm : no - tracial - states ] , very little is known about the behaviour of / on s carrying a trace . in the remainder of this paper
we suggest a possible new route to tackle this situation . [
exam : spbdd - mnc ] let @xmath194 be a linear mapping . then @xmath12 is unital , surjective and / if and only if there are a jordan automorphism @xmath195 of @xmath196 and a non - zero complex number @xmath197 such that @xmath198 where @xmath199 denotes the normalised centre - valued trace on @xmath196
. evidently , the formula ( [ equa : canonical - form ] ) defines a unital mapping which is / , since the images under @xmath195 and @xmath199 commute .
moreover , @xmath12 is surjective : let @xmath200 and @xmath201 be such that @xmath202 . put @xmath203 and @xmath204 . then @xmath205 . setting @xmath206 we have @xmath207 conversely
, if @xmath12 is spectrally bounded , it leaves @xmath208 invariant , as this space is spanned by the nilpotent matrices and @xmath12 preserves nilpotency ( @xcite , lemma 3.1 ) . assuming that @xmath12 is surjective , it is in fact bijective and
hence remains injective when restricted to @xmath208 . since the latter is finite dimensional , @xmath209 is bijective from @xmath208 to @xmath208 .
by @xcite , there exist a jordan automorphism @xmath195 of @xmath210 and a non - zero complex number @xmath197 such that @xmath211 .
hence , for each @xmath212 , @xmath213 which is nothing but identity ( [ equa : canonical - form ] ) , if @xmath12 is unital . since @xmath214 is injective , by @xcite , proposition 4.2 , it is surjective as well . by the above description ( [ equa : canonical - form ] ) ,
@xmath211 for a jordan automorphism @xmath195 of @xmath210 and @xmath215 .
suppose first that @xmath216 and take @xmath217 with @xmath218 ; for instance , @xmath219 , where @xmath220 , @xmath221 denote the standard matrix units
. then @xmath222 on the other hand , @xmath12 preserves the peripheral spectrum , by proposition 4.7 in @xcite .
therefore @xmath223 , and the claim follows from ( [ equa : canonical - form ] ) . in the case
@xmath224 , note that identity ( [ equa : canonical - form ] ) entails that @xmath225 for all @xmath212 .
since @xmath12 always preserves one eigenvalue ( in the peripheral spectrum ) , it follows that @xmath12 preserves the entire spectrum and hence must be a jordan automorphism .
the case @xmath226 is trivial .
[ prob : spsiom - ii1f ] let @xmath0 be a 1f/ with normalised centre - valued trace @xmath199 .
let @xmath227 be a unital surjective /.
are there a jordan automorphism @xmath195 of @xmath0 and a non - zero complex number @xmath197 such that @xmath228 1 .
every jordan epimorphism @xmath229 is either multiplicative or anti - multiplicative , by a classical result due to herstein , as @xmath0 is simple , hence prime .
therefore its kernel is an ideal of @xmath0 .
since @xmath0 is simple , @xmath195 must be injective . 2 . by the remark just made together with the final part of the argument in example [ exam : spbdd - mnc ] it suffices to find a jordan epimorphism @xmath195 on @xmath0 such that @xmath211 for some non - zero @xmath230 .
it is easily seen that the representation ( [ equa : canonical - form - factor ] ) is unique .
suppose that @xmath231 are two representations of the / @xmath12 , where @xmath232 and both @xmath195 , @xmath233 are jordan automorphisms of @xmath0 .
it suffices to show that @xmath234 .
since @xmath235 and both @xmath195 and @xmath233 are spectral isometries , it follows that @xmath236 .
let @xmath90 be a projection with @xmath237 and put @xmath238 .
then @xmath239 .
as both @xmath195 and @xmath233 preserve the spectrum , it follows that @xmath240 and thus @xmath234 .
suppose the / @xmath227 can be written in the form ( [ equa : canonical - form - factor ] ) .
then @xmath211 for some non - zero @xmath230 .
since @xmath241 for all @xmath242 , it follows that @xmath243 .
let @xmath90 be a projection with @xmath237 .
then @xmath244 . as in example
[ exam : spisom - mnc ] , @xmath245 .
since @xmath12 preserves the peripheral spectrum , we conclude that @xmath223 . | we discuss several open problems on spectrally bounded operators , some new , some old , adding in a few new insights .
_ dedicated to professor rainer nagel on the occasion of his retirement . _ |
the rapid development of the theory of quantum information has revived the interest in open quantum systems in connection , on one side , to the decoherence phenomenon and , on the other side , to their capacity of generating entanglement in multi - partite systems interacting with their environments .
quantum entanglement represents the physical resource in quantum information science which is indispensable for the description and performance of such tasks like teleportation , superdense coding , quantum cryptography and quantum computation @xcite .
therefore the generation , detection and manipulation of the entanglement continues to be presently a problem of intense investigation . when two systems are immersed in an environment , then , besides and at the same time with the quantum decoherence , the environment can also generate a quantum entanglement of the two systems and therefore an additional mechanism to correlate them @xcite
. in certain circumstances , the environment enhances entanglement and in others it suppresses the entanglement and the state describing the two systems becomes separable . the structure and properties of the environment may be such that not only the two systems become entangled , but also such that a certain amount of entanglement survives in the asymptotic long - time regime .
the reason is that even if not directly coupled , the two systems immersed in the same environment can interact through the environment itself and it depends on how strong this indirect interaction is with respect to the quantum decoherence , whether entanglement can be generated at the beginning of the evolution and , in the case of an affirmative answer , if it can be maintained for a definite time or it survives indefinitely in time @xcite . in this work
we study , in the framework of the theory of open quantum systems based on completely positive dynamical semigroups , the dynamics of the continuous variable entanglement for a subsystem composed of two identical harmonic oscillators interacting with its environment .
we are interested in discussing the correlation effect of the environment , therefore we assume that the two systems are independent , i.e. they do not interact directly .
the initial state of the subsystem is taken of gaussian form and the evolution under the quantum dynamical semigroup assures the preservation in time of the gaussian form of the state .
the organizing of the paper is as follows . in sect .
2 the notion of the quantum dynamical semigroup is defined using the concept of a completely positive map . then we give the general form of the kossakowski - lindblad quantum mechanical master equation describing the evolution of open quantum systems in the markovian approximation .
we mention the role of complete positivity in connection with the quantum entanglement of systems interacting with an external environment . in sec .
3 we write and solve the equations of motion in the heisenberg picture for two independent harmonic oscillators interacting with a general environment .
then , by using the peres - simon necessary and sufficient condition for separability of two - mode gaussian states @xcite , we investigate in sec .
4 the dynamics of entanglement for the considered subsystem . in particular , with the help of the asymptotic covariance matrix , we determine the behaviour of the entanglement in the limit of long times . we show that for certain classes of environments the initial state evolves asymptotically to an equilibrium state which is entangled , while for other values of the parameters describing the environment , the entanglement is suppressed and the asymptotic state is separable .
we analyze also the time evolution of the logarithmic negativity , which characterizes the degree of entanglement of the quantum state .
a summary and conclusions are given in sec .
the time evolution of a closed physical system is given by a dynamical group @xmath0 uniquely determined by its generator @xmath1 , which is the hamiltonian operator of the system .
the action of the dynamical group @xmath2 on any density matrix @xmath3 from the set @xmath4 of all density matrices in the hilbert space @xmath5 of the quantum system is defined by @xmath6 for all @xmath7 . according to von neumann ,
density operators @xmath8 are trace class ( @xmath9 ) , self - adjoint ( @xmath10 ) , positive ( @xmath11 ) operators with @xmath12 .
all these properties are conserved by the time evolution defined by @xmath2 . in the case of open quantum systems , the time evolution @xmath13 of the density operator
@xmath14 has to preserve the von neumann conditions for all times .
it follows that @xmath13 must have the following properties : @xmath15 for @xmath16 @xmath17 i. e. @xmath18 must preserve the convex structure of @xmath19 @xmath20 @xmath21 @xmath22 the time evolution @xmath23 for closed systems must be a group @xmath24 we have also @xmath25 and @xmath26 in the trace norm when @xmath27 .
the dual group @xmath28 acting on the observables @xmath29 i.e. on the bounded operators on @xmath5 , is given by @xmath30 then @xmath31 and @xmath32 , where @xmath33 is the identity operator on @xmath34 also , @xmath35 ultraweakly when @xmath27 and @xmath28 is an ultraweakly continuous mapping @xcite .
these mappings have a strong positivity property called complete positivity : @xmath36 in the axiomatic approach to the description of the evolution of open quantum systems @xcite , one supposes that the time evolution @xmath37 of open systems is not very different from the time evolution of closed systems .
the simplest dynamics @xmath37 which introduces a preferred direction in time , characteristic for dissipative processes , is that in which the group condition is replaced by the semigroup condition @xcite @xmath38 the complete positivity condition has the form : @xmath39 where @xmath40 denotes the dual of @xmath18 acting on @xmath41 and is defined by the duality condition @xmath42 then the conditions @xmath43 and @xmath44 are equivalent . also the conditions @xmath45 ultraweakly when @xmath46 and @xmath47 in the trace norm when @xmath48 are equivalent . for the semigroups with these properties and with a more weak property of positivity than eq .
( [ cposo ] ) , namely @xmath49 it is well known that there exists a ( generally unbounded ) mapping @xmath50 the generator of @xmath51 and @xmath40 is uniquely determined by @xmath52 the dual generator of the dual semigroup @xmath18 is denoted by @xmath53 : @xmath54 the evolution equations by which @xmath53 and @xmath50 determine uniquely @xmath18 and @xmath40 , respectively , are given in the schrdinger and heisenberg picture as @xmath55 and @xmath56 these equations replace in the case of open systems the von neumann - liouville equations @xmath57\end{aligned}\ ] ] and @xmath58,\end{aligned}\ ] ] respectively . for applications , eqs .
( [ meqs ] ) and ( [ meqh ] ) are only useful if the detailed structure of the generator @xmath59 is known and can be related to the concrete properties of the open systems described by such equations . for the class of dynamical semigroups which are completely positive and norm continuous , the generator @xmath50 is bounded . in many applications
the generator is unbounded . according to lindblad @xcite
, the following argument can be used to justify the complete positivity of @xmath40 : if the open system is extended in a trivial way to a larger system described in a hilbert space @xmath60 with the time evolution defined by @xmath61 then the positivity of the states of the compound system will be preserved by @xmath62 only if @xmath40 is completely positive . with this observation a new equivalent definition of the complete positivity
is obtained : @xmath40 is completely positive if @xmath62 is positive for any finite dimensional hilbert space @xmath63 the physical meaning of complete positivity can mainly be understood in relation to the existence of entangled states , the typical example being given by a vector state with a singlet - like structure that can not be written as a tensor product of vector states .
positivity property guarantees the physical consistency of evolving states of single systems , while complete positivity prevents inconsistencies in entangled composite systems and therefore the existence of entangled states makes the request of complete positivity necessary @xcite . a bounded mapping @xmath64 which satisfies @xmath65 and @xmath66 is called dissipative .
the 2-positivity property of the completely positive mapping @xmath40 : @xmath67 with equality at @xmath68 , implies that @xmath50 is dissipative .
conversely , the dissipativity of @xmath50 implies that @xmath40 is 2-positive .
@xmath50 is called completely dissipative if all trivial extensions of @xmath50 to a compound system described by @xmath60 with any finite dimensional hilbert space @xmath69 are dissipative .
there exists a one - to - one correspondence between the completely positive norm continuous semigroups @xmath40 and completely dissipative generators @xmath50 .
the following structural theorem gives the most general form of a completely dissipative mapping @xmath50 @xcite .
* theorem .
* @xmath50 is completely dissipative and ultraweakly continuous if and only if it is of the form @xmath70+\frac{1}{2\hbar } \sum_{j } ( v_{j}^\dagger[a , v_{j}]+[v_{j}^\dagger , a]v_{j}),\label{genh}\end{aligned}\ ] ] where @xmath71 . the dual generator on the state space ( schrdinger picture ) is of the form @xmath72+\frac{1}{2\hbar}\sum_{j}([v_{j}\rho , v_{j}^\dagger ] + [ v_{j},\rho v_{j}^\dagger ] ) .
\label{gens}\end{aligned}\ ] ] eqs .
( [ meqs ] ) and ( [ gens ] ) give the explicit form of the kossakowski - lindblad master equation , which is the most general time - homogeneous quantum mechanical markovian master equation with a bounded liouville operator @xcite : @xmath73 + \frac{1}{2\hbar}\sum_j([v_j\phi_t(\rho),v_j^\dagger]+[v_j,\phi_t(\rho)v_j^\dagger]).\end{aligned}\ ] ] the assumption of a semigroup dynamics is only applicable in the limit of weak coupling of the subsystem with its environment , i.e. for long relaxation times @xcite .
we mention that the majority of markovian master equations found in the literature are of this form after some rearrangement of terms , even for unbounded generators .
it is also an empirical fact for many physically interesting situations that the time evolutions @xmath18 drive the system towards a unique final state @xmath74 for all @xmath75 .
we are interested in the dynamics of entanglement in a subsystem composed of two identical non - interacting ( independent ) harmonic oscillators in weak interaction with a general environment , so that their reduced time evolution can be described by a markovian , completely positive quantum dynamical semigroup . if @xmath76 is the dynamical semigroup describing the irreversible time evolution of the open quantum system in the heisenberg picture , then the kossakowski - lindblad master equation has the following form for an operator a ( see eqs .
( [ meqh ] ) , ( [ genh ] ) ) @xcite : @xmath77 + \frac{1}{2\hbar}\sum_j(v_j^{\dagger}[\widetilde\phi_t(a ) , v_j]+[v_j^{\dagger},\widetilde\phi_t(a)]v_j).\label{masteq}\end{aligned}\ ] ] here , @xmath1 denotes the hamiltonian of the open system and the operators @xmath78 defined on the hilbert space of @xmath79 represent the interaction of the open system with the environment .
we are interested in the set of gaussian states , therefore we introduce such quantum dynamical semigroups that preserve this set . consequently @xmath1 is taken to be a polynomial of second degree in the coordinates @xmath80 and momenta @xmath81 of the two quantum oscillators and @xmath82 are taken polynomials of first degree in these canonical observables .
then in the linear space spanned by the coordinates and momenta there exist only four linearly independent operators @xmath83 @xcite : @xmath84 where @xmath85 the hamiltonian @xmath1 of the two uncoupled identical harmonic oscillators of mass @xmath86 and frequency @xmath87 is given by @xmath88 the fact that @xmath89 is a dynamical semigroup implies the positivity of the following matrix formed by the scalar products of the four vectors @xmath90 whose entries are the components @xmath91 respectively : @xmath92
its matrix elements have to be chosen appropriately to suit various physical models of the environment . for a quite general environment able to induce noise and damping effects
, we take this matrix of the following form , where all the coefficients @xmath93 ... and @xmath94 are real quantities , representing the diffusion coefficients and , respectively , the dissipation constant : @xmath95 it follows that the principal minors of this matrix are positive or zero . from the cauchy - schwarz inequality
the following relations for the coefficients defined in eq .
( [ coef ] ) hold ( from now on we put , for simplicity , @xmath96 ) : @xmath97 the matrix of the coefficients ( [ coef ] ) can be conveniently written as ( @xmath98 denotes the transposed matrix ) @xmath99 in terms of @xmath100 matrices @xmath101 , @xmath102 and @xmath103 .
this decomposition has a direct physical interpretation : the elements containing the diagonal contributions @xmath104 and @xmath105 represent diffusion and dissipation coefficients corresponding to the first , respectively the second , system in absence of the other , while the elements in @xmath106 represent environment generated couplings between the two oscillators , taken initially independent .
we introduce the following @xmath107 bimodal covariance matrix : @xmath108 with the correlations of operators @xmath109 and @xmath110 defined by using the density operator @xmath3 of the initial state of the quantum system , as follows : @xmath111 by using eq .
( [ masteq ] ) we obtain by direct calculation the following systems of equations for the quantum correlations of the canonical observables @xcite : @xmath112 where @xmath113 @xmath114 introducing the notation @xmath115 the time - dependent solution of eq .
( [ vareq ] ) is given by @xcite @xmath116 where the matrix @xmath117 has to fulfill the condition @xmath118 in order that this limit exists , @xmath119 must only have eigenvalues with negative real parts .
the values at infinity are obtained from the equation @xmath120
the two - mode gaussian state is entirely specified by its covariance matrix ( [ covar ] ) , which is a real , symmetric and positive matrix with the following block structure : @xmath121 where @xmath122 , @xmath123 and @xmath124 are @xmath100 matrices .
their entries are correlations of the canonical operators @xmath125 and @xmath126 @xmath122 and @xmath123 denote the symmetric covariance matrices for the individual reduced one - mode states , while the matrix @xmath124 contains the cross - correlations between modes .
the elements of the covariance matrix depend on @xmath119 and @xmath127 and can be calculated from eqs .
( [ covart ] ) , ( [ covarinf ] ) . since the two oscillators are identical , it is natural to consider environments for which the two diagonal submatrices in eq .
( [ subm ] ) are equal , @xmath128 and the matrix @xmath106 is symmetric , so that in the following we take @xmath129 then both unimodal covariance matrices are equal , @xmath130 and the entanglement matrix @xmath124 is symmetric .
it is interesting that the general theory of open quantum systems allows couplings via the environment between uncoupled oscillators .
according to the definitions of the environment parameters , the diffusion coefficients above can take non - zero values and therefore can simulate an interaction between the uncoupled oscillators .
consequently , the cross - correlations between modes can have non - zero values . in this case
the gaussian states with @xmath131 are separable states , but for @xmath132 it may be possible that the states are entangled , as will be shown next . in order to investigate whether an external environment can actually entangle the two independent systems , we use the partial transposition criterion @xcite : a state is entangled if and only if the operation of partial transposition does not preserve its positivity .
for the particular case of gaussian states , simon @xcite obtained the following necessary and sufficient criterion for separability : @xmath133 where @xmath134- \frac{1}{4}(\det a+\det b ) \label{sim1}\end{aligned}\ ] ] and @xmath135 is the @xmath100 symplectic matrix @xmath136 in the following we consider such environment diffusion coefficients , for which @xmath137 this corresponds to the case when the asymptotic state is a gibbs state @xcite . in order to describe the dynamics of entanglement
, we have to analyze the time evolution of the simon function @xmath138 ( [ sim1 ] ) .
we consider two cases , according to the type of the initial gaussian state : separable or entangled .
\1 ) to illustrate a possible generation of the entanglement , we represent in figure 1 the function @xmath138 versus time @xmath139 and diffusion coefficient @xmath140 for a separable initial gaussian state with initial correlations @xmath141 we notice that , according to peres - simon criterion , for relatively small values of the coefficient @xmath142 the initial separable state remains separable for all times . for larger values of @xmath142 at some finite moment of time ,
when @xmath138 becomes negative , the state becomes entangled . in some cases
the entanglement is only temporarily generated , that is the state becomes again separable .
after that , at a certain moment of time , one can notice again a revival of entanglement .
in these cases the generated entangled state remains entangled forever , including the asymptotic final state . versus time @xmath139 and environment coefficient @xmath142 for @xmath143 @xmath144 and a separable initial gaussian state with initial correlations @xmath145 we have taken @xmath146 , title="fig : " ] \2 ) the evolution of an entangled initial state is illustrated in figure 2 , where we represent the function @xmath138 versus time and diffusion coefficient @xmath147 for an initial entangled gaussian state with initial correlations @xmath148 we notice that for relatively small values of @xmath142 at some finite moment of time @xmath138 takes non - negative values and therefore the state becomes separable .
this is the so - called phenomenon of entanglement sudden death .
depending on the values of the coefficient @xmath142 it is also possible to have a repeated collapse and revival of the entanglement .
one can also show that for relatively large values of the coefficients @xmath149 and @xmath142 the initial entangled state remains entangled for all times .
, title="fig : " ] on general grounds , one expects that the effects of decoherence , counteracting entanglement production , is dominant in the long - time regime , so that no quantum correlation ( entanglement ) is expected to be left at infinity . nevertheless , there are situations in which the environment allows the existence of entangled asymptotic equilibrium states . from eq .
( [ covarinf ] ) we obtain the following elements of the asymptotic entanglement matrix @xmath150 : @xmath151 @xmath152 @xmath153 the elements of matrices @xmath154 and @xmath155 are obtained by putting @xmath156 in the previous expressions .
we calculate the determinant of the entanglement matrix and obtain : @xmath157.\end{aligned}\ ] ] with the chosen coefficients ( [ envcoe ] ) , the simon expression ( [ sim1 ] ) takes the following form in the limit of large times : @xmath158 for environments characterized by such coefficients that the expression @xmath159 ( [ sim2 ] ) is strictly negative , the asymptotic final state is entangled . just to give an example , without altering the general features of the system , we consider the particular case of @xmath160 then we obtain that @xmath161 i.e. the asymptotic final state is entangled , for the following range of values of the coefficient @xmath162 characterizing the environment @xcite : @xmath163 where the diffusion coefficient @xmath149 satisfies the condition @xmath164 equivalent with the unimodal uncertainty relation .
we remind that , according to inequalities ( [ coefineq ] ) , the coefficients have to fulfill also the constraint @xmath165 if the coefficients do not fulfil the inequalities ( [ insep ] ) , then @xmath166 and the asymptotic state of the considered system is separable . these results show that , irrespective of the initial conditions , we can obtain either an separable or an inseparable asymptotic entangled state , for a suitable choice of the diffusion and dissipation coefficients .
versus time @xmath139 and environment coefficient @xmath142 for @xmath143 @xmath144 and a separable initial gaussian state with initial correlations @xmath145 we have taken @xmath146 , title="fig : " ] we apply the measure of entanglement based on negative eigenvalues of the partial transpose of the subsystem density matrix . for a gaussian density operator ,
the negativity is completely defined by the symplectic spectrum of the partial transpose of the covariance matrix . the logarithmic negativity @xmath167 $ ] determines the strength of entanglement for @xmath168 if @xmath169
then the state is separable . here
@xmath170 ^ 2-\det\sigma(t)}\right)^{1/2}.\end{aligned}\ ] ] in figures 3 and 4
we represent the logarithmic negativity @xmath171 versus time @xmath139 and diffusion coefficient @xmath140 for the two types of the initial gaussian state , separable or entangled , with the same initial correlations , previously considered when we analyzed the time evolution of the simon function @xmath172 as expected , we remark that the logarithmic negativity has a behaviour similar to that one of the simon function in what concerns the characteristics of the state of being separable or entangled . depending on the values of the environment coefficients , the initial state can preserve for all times its initial property
separable or entangled , but we can also notice the generation of entanglement or the collapse of entanglement ( entanglement sudden death ) at those finite moments of time when the logarithmic negativity @xmath171 reaches zero value .
one can also observe a repeated collapse and revival of the entanglement . in the case of an entangled initial state ,
the logarithmic negativity is a fluctuating function and decreases asymptotically in time .
, title="fig : " ] in our case the asymptotic logarithmic negativity has the form @xmath173.\label{aslog}\end{aligned}\ ] ] it depends only on the diffusion and dissipation coefficients characterizing the environment and does not depend on the initial gaussian state .
one can easily see that the double inequality ( [ insep ] ) , determining the existence of asymptotic entangled states ( @xmath174 ) is equivalent with the condition of positivity of the expression ( [ aslog ] ) of the logarithmic negativity , @xmath175
we have given a brief review of the theory of open quantum systems based on completely positive quantum dynamical semigroups and mentioned the necessity of the complete positivity for the existence of entangled states of systems interacting with an external environment .
in the framework of this theory we investigated the dynamics of the quantum entanglement for a subsystem composed of two uncoupled identical harmonic oscillators interacting with a common environment . by using the peres - simon necessary and sufficient condition for separability of two - mode gaussian states , we have described the generation and evolution of entanglement in terms of the covariance matrix for an arbitrary gaussian input state . for some values of diffusion and dissipation coefficients describing the environment ,
the state keeps for all times its initial type : separable or entangled . in other cases , entanglement generation or entanglement suppression ( entanglement sudden death ) take place or even one can notice a repeated collapse and revival of entanglement . we have also shown that , independent of the type of the initial state , for certain classes of environments the initial state evolves asymptotically to an equilibrium state which is entangled , while for other values of the coefficients describing the environment , the asymptotic state is separable .
we described also the time evolution of the logarithmic negativity , which characterizes the degree of entanglement of the quantum state .
the existence of quantum correlations between the two considered harmonic oscillators interacting with a common environment is the result of the competition between entanglement and quantum decoherence . from the formal point of view , the generation of entanglement or its suppression ( entanglement sudden death ) correspond to the finite time vanishing of the simon separability function or , respectively , of the logarithmic negativity .
presently there is a large debate relative to the physical interpretation existing behind these fascinating phenomena . due to the increased interest manifested towards the continuous variables approach @xcite to quantum information theory
, these results , in particular the possibility of maintaining a bipartite entanglement in a diffusive - dissipative environment for asymptotic long times , might be useful in controlling the entanglement in open systems and also for applications in the field of quantum information processing and communication .
the author acknowledges the financial support received within the project ceex 68/2005 . | in the framework of the theory of open systems based on completely positive quantum dynamical semigroups , we study the continuous variable entanglement for a system consisting of two independent harmonic oscillators interacting with a general environment .
we solve the kossakowski - lindblad master equation for the time evolution of the considered system and describe the entanglement in terms of the covariance matrix for an arbitrary gaussian input state . using peres
simon necessary and sufficient criterion for separability of two - mode gaussian states , we show that for certain values of diffusion and dissipation coefficients describing the environment , the state keeps for all times its initial type : separable or entangled . in other cases , entanglement generation , entanglement sudden death or a periodic collapse and revival of entanglement take place .
we analyze also the time evolution of the logarithmic negativity , which characterizes the degree of entanglement of the quantum state . |
arp 220 ( ic 4553/4 ) , with an infrared luminosity of 1.5 x 10@xmath3 at @xmath4=8 - 1000 @xmath0 m , is one of the nearest ultraluminous infrared galaxies ( @xcite ) .
visual wavelength images reveal two faint tidal tails , indicating a recent tidal interaction ( cf .
@xcite ) , and high resolution ground - based radio and near - infrared imaging show a double nucleus ( @xcite , @xcite ) .
the radio nuclei are separated by 0.98 at pa @xmath5 ( @xcite ) .
the system is also extraordinarily rich in molecular gas ( @xmath6 , scoville , yun & bryant 1997 ) and about 2/3 of this gas and dust is concentrated in a thin disk with radii @xmath7 pc .
ground - based observations at 10 to 30 @xmath0 m suggest that the far - infrared luminosity originates from a region of similar size ( @xcite ) . to power
the energy output seen in the infrared by young stars requires a star formation rate @xmath8 yr@xmath9 .
alternatively , if the luminosity originates from an agn , this source must be sufficiently obscured by dust that even the mid - infrared emission lines are highly extinguished since spectroscopy with iso shows no evidence of very high ionization lines at wavelengths out to 40@xmath0 m ( @xcite ) . in this letter
we report near infrared imaging at @xmath4=1 - 2.2@xmath0 m with 0.10.2 resolution obtained using the nicmos camera on the hst .
these high resolution data clearly reveal the complex nature of the nuclei with the morphologies of both near - infrared nuclei strongly affected by dust obscuration and enable an improved registration of the infrared relative to the radio nuclei . throughout this letter
, we adopt a distance of 77 mpc ( 1 corresponds to 373 pc ) for arp 220 using h@xmath10=75 mpc@xmath9 ( @xcite ) .
arp 220 was observed using the nicmos camera 2 on the hst on 4 april 1997 as part of the early release observations ( ero ) program .
camera 2 has a 256@xmath11256 hgcdte array with 0.076 pixels , providing a 19.2 field of view .
images were obtained in the f110w , f160w and f222 m filters , providing resolutions of 0.11 , 0.16 and 0.22 respectively . for both the galaxy and an hst guide star , observed to determine the point - spread - function ( psf ) , a four step dither pattern was executed with offsets of 5.4 pixels ( 0.41 ) . at each position in the dither pattern on arp 220 ,
two integrations , each lasting 128 sec , with multiple non - destructive reads ( multiaccum ) were obtained .
the data were dark subtracted , flat fielded and corrected for cosmic rays using the iraf software nicmos - pipeline routines .
the dithered images were shifted and coadded using the drizzle routine in iraf .
the flux scale employed calibration factors of 2.35@xmath11 , 2.77@xmath11 , and 6.11@xmath12 jy ( adu / sec)@xmath9 at 1.1 , 1.6 , and 2.22 @xmath0 m ( @xcite ) .
the measured fluxes at h ( 1.6@xmath0 m ) and k ( 2.22@xmath0 m ) for the nuclear sources and for a 5 aperture agree to better than 10% with those obtained from ground - based imaging by carico ( 1990 ) . at 1.1@xmath0 m
a direct comparison is not possible since the standard j filter is at 1.25@xmath0 m .
the rms noise levels in the final images are 12 , 15 , 35 @xmath0jy ( arcsec)@xmath13 at 1.1 , 1.6 , and 2.22 @xmath0 m , respectively . for observations with nicmos camera 2 ,
the angular resolutions are a factor of two lower at 2.22@xmath0 m than at 1.1@xmath0 m .
this can seriously hinder attempts to analyze color gradients where there are compact emission sources unless the images at each wavelength are convolved or deconvolved to a common resolution .
we have adopted two approaches : convolving the two short wavelength images to the resolution of the raw image at 2.22@xmath0 m and deconvolving the two long wavelength images to the resolution at 1.1@xmath0 m .
the former is , of course , the safest approach but provides more limited spatial resolution . for the latter approach
, deconvolution was done using the richardson - lucy algorithm without damping ( @xcite ) to achieve 0.14 resolution . in figure 1 ( color plate ) the images for the entire camera 2 field from the 3 bands are combined in false color .
the image clearly reveals the structure around the twin nuclei separated by @xmath1 1 e - w in addition to extended emission from the system out to @xmath148 .
at least eight unresolved ( @xmath15pc ) sources are also seen at radii 27 the brightest being approximately 5 southwest of the nucleus .
all of them have bluer colors than the nuclei but their magnitudes are sufficiently bright that they can not be individual stars .
we hereafter refer to them as clusters .
the brightest of these clusters ( 1 , 2 , and 4 ) are seen in the visible band hst imaging of shaya ( 1994 ) and they are also clearly seen in subsequent i - band hst imaging ( borne and lucas 1997 ) . the central region is shown in contour form in figures 2 - 4 ( plates 2 - 4 ) using : the original drizzled images at each wavelength ; images convolved to 0.22@xmath16 resolution ; and deconvolved to 0.11@xmath16 resolution , respectively . in each case , the lower right panel shows the ratio of the 2.22@xmath0m/1.1@xmath0 m data .
coordinate offsets in figures 2 - 4 are measured from the peak of the deconvolved 2.22@xmath0 m emission in the west nucleus .
flux measurements for the nuclear components and all identified cluster sources are listed in table 1 along with measured sizes ( fwhm ) for resolved structures .
most of the near infrared emission from the nuclear region of arp220 must be stellar ( rather than hot dust or shock - excited emission lines ) since the strength of the co overtone absorption at 2.35@xmath0 m requires that @xmath1490% of the observed near - infrared flux in a 5 aperture arises from stars ( @xcite , @xcite ) . in previous ground - based images ( eg .
@xcite ) , the core of arp 220 appears as two nuclei separated e - w by 0.95 . in the hst - nicmos images , the eastern nucleus breaks up into northern ( ne ) and southern ( se ) components separated by 0.4 the latter appears highly reddened and is barely visible in the 1.1@xmath0 m image . and
, the bright nucleus on the west ( w ) , is seen as a _ crescent _ arc in the new images .
thus , for both nuclei , their morphology , even at @xmath4=2.2@xmath0 m ( where the extinction is 10 times less than a@xmath17 ) appears strongly influenced by dust extinction . in addition , a new component is seen @xmath18 south of the western nucleus ( s ) and another midway between the two nuclei .
the accuracy of the guide star position for the arp 220 observations was insufficient to independently determine the absolute positions of the near infrared images .
we have therefore registered the infrared images by comparison with cm - wave radio continuum maps for which the coordinates are well - determined .
graham ( 1990 ) matched the brightest radio sources ( a and b , @xcite ) with the brightest portions of the eastern and western nuclei ( i.e. components ne and w ) .
however , this placement now seems unlikely since the separation and position angles of the two radio components and the two brightest infrared components are significantly different .
instead , we suggest that both of the two brightest radio components are located within the areas of very high extinction south of the w component and between the ne and se sources .
this registration places the weak radio source c precisely on the southern ( s ) infrared source which is relatively blue and therefore unlikely to be strongly extincted . using the 2.22@xmath0m/1.1@xmath0 m ratio image ( figure 3 ) to estimate the relative extinction across the region ,
we have crudely corrected the 2.22@xmath0 m image for extinction under the assumption that the dust is in the foreground , i.e. not mixed with the stars . in figure 5 ,
the radio source contours from baan & haschick ( 1995 ) are shown superposed on this ` extinction - corrected ' 2.22@xmath0 m image using the registration discussed above .
radio source b lies within the dark area below the crescent on the w source while a is at the north of the reddened se infrared source .
support for our registration of the near infrared emission is provided by the strong morphological similarity of the infrared structure shown here and the h@xmath2co emission ( @xcite ) the latter exhibits similar offsets relative to the radio continuum .
the 2.22@xmath0m/1.1@xmath0 m ratio images in figures 2 - 4 may be used to probe variations in the extinction assuming that the intrinsic color of the underlying stars is not strongly varying .
the false color image for the deconvolved data ( plate 1 ) clearly shows an overall increase in the reddening towards the central regions of arp 220 in the annulus at 2 - 4 radii , m@xmath19=2.30 mag , while inside 2 radius , m@xmath19=3.19 mag . in the central 2 , figure 3 reveals even larger variations : high reddening is seen over the se source where the color ( 1.1 - 2.2@xmath0 m ) is 3.86 mag and in a nearly complete ring surrounding the bright w source where the color is 3.31 mag . for a population of stars with ages @xmath20 yrs
, typical colors are j - h = -0.2 , h - k @xmath10.0 mag . for a standard extinction law ( @xcite ) , and the young population colors , the derived visual extinctions are a@xmath17= 17 ( w ) , 15(s ) , 24(se ) , 18(ne ) , and 13(all ) mag based on the observed h - k magnitudes given in table 1 . for an old population ( j - h @xmath10.74 , h - k @xmath10.24 mag ) , the visual extinctions are 13 , 11 , 20 , 14 , and 9 mag respectively , similar to the 1 resolution estimates obtained by larkin ( 1995 ) from pa@xmath21 and br@xmath22 line ratios .
considerably higher estimates were obtained by sturm ( 1996 ) and it is clear that there are undoubtedly areas in the nucleus from which virtually no radiation escapes , even at 2.2@xmath0 m .
the complex near infrared structure of arp 220 is due to there being multiple centers of star formation activity ( as in arp 299 , cf .
wynn - williams 1991 ) and strongly varying dust obscuration within the merger nuclei .
the relatively simple double - nuclei structure of the radio continuum ( @xcite ) and the mm - wave dust continuum ( @xcite ) suggest that the major mass concentrations are associated with the w and se sources .
the ne and s sources would then be relatively minor mass concentrations , albeit with high present - epoch luminosity .
if the intrinsic , _ unobscured _ shapes of the expected stellar emission are either ellipsoidal ( in the case of the nuclear core clusters ) , point - like ( for unresolved stellar clusters ) or disk - like ( for young stars formed in disk ) , the unobscured morphology should appear spherical or elliptical depending on the viewing angle .
for both the e and w nuclei , the morphology at 1.1 - 2.2@xmath0 m is clearly not elliptical : the western nucleus appears as a crescent and the eastern nucleus is split into two components ( ne and se ) with very different colors .
given our suggested registration of the radio and infrared , it is evident that the true centers of the nuclei are so obscured that neither is seen directly at 2.2@xmath0 m .
the crescent or partial - ring morphology of the western component is clearest in the deconvolved 2.2@xmath0 m image ( figure 4 ) , but it is also apparent in the original 2.2@xmath0 m data ( figure 2 ) .
this morphology might readily arise under two circumstances : if there is an obscuring disk of dust ( and gas ) embedded within the _ spheroidal _ nuclear star cluster or if a central starburst _ ring _ or disk of young stars is partially obscured by its own dust .
figure 6 models the surface brightness distribution for a star cluster in which the midplane is cut by a totally opaque dust disk ; here , the observed crescent - shape arises due to an increase in geometrical path length through the cluster periphery . the second hypothesis with a ring or torus of massive star formation is similar to models proposed for seyfert galaxy nuclei ( eg .
maolino & rieke 1995 , genzel 1995 ) .
two important characteristics of the observed emission should be emphasized : the crescent shape and the extremely abrupt cutoff by dust to the south of the w source .
the former strongly implies that the dust is _ embedded _ in the core cluster / disk rather than being a foreground dust lane ; the latter implies that the dust is in a extremely thin , but very opaque disk .
this geometry agrees with that suggested by scoville , yun & bryant ( 1997 ) based on arcsec resolution imaging of the mm - co line .
the low velocity dispersion of the molecular gas implied a very thin disk ( @xmath23z=16 pc ) .
a similar disk structure in the nuclei has been suggested by baan & haschick ( 1995 ) based on their h@xmath2co data .
the morphology of the eastern nucleus is not so distinctive , yet the apparent bifurcation of the nucleus into lowly and highly reddened components ( ne and se respectively ) and the probable positioning of the radio nucleus in an area of weak 2.2@xmath0 m emission both suggest a nuclear dust disk like that in the w nucleus . for both nuclei ,
the highest extinctions occur to the south of the radio centers , suggesting that the dust disks are tilted such that the south is the near side . the measured infrared sizes ( fwhm ) of the w , se and ne source components are 0.2 - 0.5 , corresponding to 75 - 185 pc .
the separation of the radio nuclei ( a and b ) is 0.98 or 364 pc in projection .
their actual separation is @xmath1600 pc if the geometry suggested by baan & haschick ( 1995 ) and scoville , yun & bryant ( 1997 ) is correct since the nuclei would then be well off the major axis of their orbits . within each of the nuclei , scoville , yun & bryant ( 1997 ) detected co emission profiles of full - width @xmath1250 .
if the rotation velocities of gas within each nuclear cluster is @xmath1125 at 75 pc radius , their dynamical masses are @xmath24 .
magnitudes and positions of eight clusters for which reliable measurements could be made are listed in table 1 in order of decreasing flux at 1.1 @xmath0 m .
all of these sources are too bright to be individual stars . given our resolution ( @xmath1 30 pc ) , the fact that they are unresolved is entirely consistent with their being globular clusters .
their measured offsets from the 2.2@xmath0 m centroid imply projected galactic radii of 0.92.7 kpc .
the apparent clustering of these sources around arp 220 , with none seen beyond 7.5 radius , suggests that most of them must be associated with arp 220 rather than in the foreground .
given the fact that we have measured only two colors and must use one color to constrain the reddening , the remaining color does not strongly discriminate the ages of these clusters .
the cluster sources outside 2 radius have measured colors at 1.1 , 1.6 , and 2.2@xmath0 m which are consistent ( within the uncertainties ) with those of globular clusters in ngc 5128 ( @xcite ) , indicating that they could have age @xmath25 yrs and essentially no near - infrared reddening ( a@xmath26 5mag ) .
the absolute k magnitude of the brightest cluster ( m@xmath27= -13.5 mag ) in arp 220 is @xmath11.5 magnitudes brighter than the brightest cluster in ngc 5128 ( @xcite ) and is comparable in brightness to the tip of the globular cluster luminosity function for ellipticals ( @xcite ) .
very young much that massive , globular clusters event in the arp 220 ( cf .
@xcite ) .
the nicmos project has been supported by nasa grant nag 5 - 3042 .
it is a pleasure to acknowledge assistance and discussions of l. armus , l. bergeron , c. boone , j. goldader , a. fruchter , j. jensen , t. soifer , j. surace , and r. white .
we are grateful to k. borne and r. lucas for sharing their i - band image of arp 220 with us .
lrrcccccccc w & 0.00 & 0.00 & 16.29 & 15.40 & 14.10 & 13.04 & 1.30 & 1.06 & @xmath28 & @xmath29 s & e 0.21 & @xmath30 & 18.00 & 17.28 & 16.09 & 15.16 & 1.19 & 0.93 & @xmath31 & @xmath32 se & e 1.00 & @xmath33 & 18.70 & 17.68 & 16.31 & 14.83 & 1.37 & 1.48 & @xmath34 & @xmath32 ne & e 1.13 & 0.07 & 17.47 & 16.57 & 15.27 & 14.14 & 1.30 & 1.13 & @xmath35 & @xmath36 all sources & e 0.57 & @xmath37 & 13.72 & 13.08 & 11.95 & 11.13 & 1.13 & 0.82 & & 5circular clusters : & & & & & & & & & & 1 & w 1.47 & @xmath38 & 20.50 & 20.15 & 19.34 & 19.12 & 0.81 & 0.22 & 37,43,56 & 46,54,70 2 & e 4.76 & @xmath39 & 21.35 & 21.03 & 20.25 & 20.21 & 0.78 & 0.04 & & 3 & w 0.63 & @xmath40 & 21.71 & 21.16 & 20.11 & & 1.05 & & & 4 & w 5.17 & 4.56 & 21.81 & 21.56 & 20.89 & & 0.66 & & & 5 & e 1.29 & @xmath41 & 22.14 & 21.70 & 20.77 & 20.70 & 0.93 & 0.07 & & 6 & e 3.83 & @xmath42 & 22.48 & 22.19 & 21.47 & & 0.72 & & & 7 & w 0.58 & 4.74 & 22.87 & 22.52 & 21.71 & & 0.81 & & & 8 & e 7.11 & 1.83 & 22.89 & 22.32 & 21.25 & & 1.07 & & & -6 mm | we report high resolution imaging of the ultraluminous infrared galaxy arp 220 at 1.1 , 1.6 , and 2.22 @xmath0 m with nicmos on the hst .
the diffraction - limited images at 0.10.2 resolution clearly resolve both nuclei of the merging galaxy system and reveal for the first time a number of luminous star clusters in the circumnuclear envelope .
the morphologies of both nuclei are strongly affected by dust obscuration , even at 2.2 @xmath0 m : the primary nucleus ( west ) presents a crescent shape , concave to the south and the secondary ( eastern ) nucleus is bifurcated by a dust lane with the southern component being very reddened . in the western nucleus ,
the morphology of the 2.2 @xmath0 m emission is most likely the result of obscuration by an opaque disk _ embedded _ within the nuclear star cluster .
the morphology of the central starburst - cluster in the western nucleus is consistent with either a circumnuclear ring of star formation or a spherical cluster with the bottom half obscured by the embedded dust disk .
comparison of cm - wave radio continuum maps with the near - infrared images suggests that the radio nuclei lie in the dust disk on the west and near the highly reddened southern component of the eastern complex .
the radio nuclei are separated by 0.98 ( corresponding to 364 pc at 77 mpc ) and the half - widths of the infrared nuclei are @xmath10.2 - 0.5 .
at least 8 , unresolved infrared sources probably globular clusters are also seen in the circumnuclear envelope at radii 2 - 7 .
their near - infrared colors do not significantly constrain their ages .
= = = 1=1=0pt = 2=2=0pt 2 h@xmath2 = = = 1=1=0pt = 2=2=0pt |
relativistic shells have found applications in an impressive list of different scenarios , such as sources of vacuum gravitational fields @xcite-@xcite , cosmological models @xcite and @xcite , gravitational collapse @xcite , canonical quantization of gravity @xcite and @xcite , higher dimensional models in brane world , quantum gravitational collapse @xcite , entropy of black holes , motion of thin clouds surrounding an exploding star @xcite , sources of phantom " fields and dark energy @xcite . in spite of the wide range of their applications , very few analytical solutions in closed form are known even in spherical symmetry .
exceptions are the dust ( i.e. zero pressure ) shells , which were completely analyzed already in the pioneering work by israel @xcite , @xcite , and solutions with linear - barotropic equations of state of the form @xmath0 with a constant @xmath1 . in the present paper
we discuss the general exact solution for shells satisfying an equation of state in closed form @xmath2 with some general assumption of physical reasonableness on the behavior of the function @xmath3 .
our solutions include as particular cases the linear e.o.s .
cited above but also , for instance , non - linear polytropic equations of state of the form @xmath4 @xcite .
it turns out that the qualitative behavior of spherical shells always presents two branches " , an
astro - physically relevant " branch in which the weak energy condition is strictly satisfied , and a phantom " branch in which the density remains positive but the active gravitational mass @xmath5 is strictly negative .
we consider a spherical shell in vacuum .
the shell separates " a portion of minkowski space ( interior " ) from schwarzschild space of arbitrarily fixed mass parameter @xmath6 ( exterior " ) .
the field equations relate the jump of the exterior curvature with the matter content of the shell , i.e. a surface distribution of stress - energy .
it can be shown @xcite that such equations can be written as follows : @xmath7 @xmath8 where @xmath9 is the radius " of the shell as a function of the proper time on the shell , @xmath10 and @xmath11 are the surface ( rest - frame ) energy - density and the surface pressure ( due to spherical symmetry , the two eigenvalues of the stress are equal , i.e. the material is necessarily a two - dimensional perfect fluid " ) and the effective potential " @xmath12 is given by @xmath13 the equations ( [ conser ] ) and ( [ rpunto ] ) contain three unknowns , and become a closed set once a further relation has been given .
this has been achieved in various ways in the literature ; for instance , one can assign the pressure profile as a a priori " known function @xcite and calculate the corresponding density profile .
however , in this way , the resulting equation of state " , i.e. the relationship between density and pressure which holds during the dynamics may turn out to be unphysical . here
, we are going to consider the case of a closed equation of state of the barotropic type , namely @xmath14 . in this case , as we shall see , the equations above decouple and the dynamics can be qualitatively analyzed .
the function @xmath15 has to satisfy a certain set of assumptions which assure the physical reasonability of the model .
it seems that a reasonable compromise between generality and physical reasonableness can be obtained assuming that @xmath16 is a continuous , monotonic function , differentiable except perhaps at @xmath17 , where its value vanishes .
these assumptions comprise a wide range of equations of state and , in particular , the linear - barotropic ( @xmath18 with a constant @xmath19 ) that has been widely studied in the literature @xcite , @xcite , and the polytropic case @xmath20 ( both these e.o.s .
reduce to the newtonian polytropes in the weak field regime @xcite , @xcite ) .
if the weak energy condition is required , then solutions have to be selected in such a way that @xmath21 , and on such solutions the effective gravitational mass @xmath22 must be non negative
. therefore the requirement @xmath23 has to be added .
however , shells with positive density but negative effective mass are of interest on their own , for instance as toy models for sources of dark energy ( `` phantom '' fields , see e.g. @xcite , exotic matter and wormholes , see @xcite ) , thus we shall consider also this possibility in what follows .
we take initial data @xmath24 and consider the behavior of the solutions of the ordinary differential equation @xmath25 first of all , we notice that it can be reduced to a quadrature @xmath26 thus , if the function @xmath27 is integrable in neighborhood of @xmath28 , the density diverges then this occurs at a finite non zero value @xmath29 ( with @xmath30 ) ( case a " ) . if instead the function @xmath27 is not integrable in neighborhood of @xmath28 , the density can diverge only as @xmath31 approaches zero ( case b " ) .
further , a straightforward analysis shows that the following cases have to be distinguished : + case i ) : @xmath32 for any @xmath10 : in this case each solution is strictly decreasing from @xmath28 to zero ( one among behaviors a or b occurs ) .
weak energy condition is always satisfied .
since @xmath17 is a solution , if @xmath33 is continuous in @xmath17 the standard uniqueness theorem for ode gives @xmath34 as @xmath35 .
for completeness we notice , however , that one could also conceive models in which @xmath33 has a finite jump in @xmath17 . in this case
a loss of uniqueness occurs at a finite value @xmath36 ( at which the solution goes to zero and can be prolonged to be zero , thus giving a sort of evaporation " of the shell ) .
we thus have four qualitatively different cases reported in figures 1 - 4 .
+ case ii ) : there exist a zero ( at @xmath37 ) of the function @xmath38 : clearly @xmath39 is a regular solution at which no loss of uniqueness can occur .
thus the behavior depends on the initial datum for @xmath10 : ( iia ) if @xmath40 then all solutions are strictly decreasing and approach from above @xmath37 as @xmath35 ; one among cases a and b occurs .
weak energy condition is satisfied everywhere .
( iib ) if @xmath41 then all solutions are strictly increasing and approach @xmath37 from below as @xmath35 ; their behavior in the past " intersects @xmath17 where a loss of uniqueness occurs .
weak energy condition is violated everywhere because the density is positive but @xmath42 is strictly negative .
two behaviors must still be distinguished : if the derivative @xmath43 has no zeroes , the solutions come out from zero with infinite tangent , if it has a zero the solutions have a flex point , and the matching at @xmath17 is smooth in this case .
all in all we have the four qualitatively different cases reported in figures 5 - 8 .
once the behavior of the energy function @xmath44 is known the corresponding qualitative behavior of the shell radius @xmath9 can be read off from the effective potential " equation ( [ rpunto ] ) .
first of all , we notice that , due to the presence of the @xmath45 term , the potential always diverges at the left end side of the interval of definition , in all the eight qualitatively different cases discussed above .
the behavior at the other extreme is divergent as well in the cases i without loss of uniqueness and in all cases ii ( where @xmath12 diverges as @xmath31 goes to infinity ) .
thus in all such cases the potential is a continuous curve positively diverging at the extremes , and it follows that it has an absolute minimum .
the dynamics is allowed in both the neighborhoods of the extremes , and it is allowed everywhere if @xmath12 is positive at the minimum .
bouncing or also oscillating behaviors can appear as well , if the minimum is negative and therefore there is at least one region in which the dynamics is not allowed .
shells with general equation of state can thus be viewed as very useful
toy " models for astrophysical objects , in order to investigate the effect of the equation of state on the dynamics in several physically interesting scenarios .
work in this direction is in progress .
berezin , v.a .
kuzmin and i.i .
tkachev , _ phys .
d _ * 36 * , 2919 ( 1987 ) .
v. dela cruz and w. israel , _
nuovo cimento _ * 51 a * , 744 ( 1967 ) .
w. israel _ phys rev .
d _ * 2 * , 641 ( 1970 ) .
v. dela cruz and w. israel , _ phys .
rev . _ * 170 * , 1187 ( 1967 ) .
h. sato _ prog .
phys . _ * 76 * , 1250 ( 1986 ) . k. lake _ ap . j. _ * li * , 401 ( 1992 ) .
l. lindblom and d.r .
d _ * 10 * , 3151 ( 1974 ) .
j. kijowski , gen .
* 29 * , 307 ( 1997 ) j. kijowski , acta phys .
b. * 29 * , 1001 ( 1998 ) p. hajicek , b.s .
kay and k. kuchar _ phys .
d _ * 46 * , 5439 ( 1992 ) .
d. nunez , arxiv : gr - qc/9710110 ( 1997 ) .
v. berezin , v. dokuchaev , yu . eroshenko and a. smirnov , arxiv : gr - qc/0503068 v1 ( 2005 ) .
w. israel , _
nuovo cimento _ * 44b * , 1 ( 1966 ) ( erratum * 48b * 463 ) .
w. israel , _ phys .
rev . _ * 153 * , 1388 ( 1967 ) .
j. kijowski , g. magli and d. malafarina , relativistic dynamics of spherical timelike shells , preprint .
d. nunez and h.p .
deoliveira , _ phys .
lett . a _ * 214 * , 227 ( 1996 ) .
zloshchastiev , arxiv : gr - qc/9802041 v1 ( 1999 ) .
goncalves , arxiv : gr - qc/0212124 v1 ( 2002 ) .
tooper , ap .
j. _ _ * 140 * , 434 ( 1964 ) .
tooper , ap .
j. _ _ * 142 * , 1541 ( 1965 ) .
m. visser _ nucl .
b _ * 328 * , 203 ( 1989 )
. francisco s. n. lobo _ class.quant.grav . _
* 21 * ( 2004 ) 4811 - 4832 ; ernesto f. eiroa , claudio simeone _ phys.rev .
d _ * 70 * ( 2004 ) 044008 ; ernesto f. eiroa , claudio simeone _ phys.rev .
d _ * 71 * ( 2005 ) 127501 . | the general exact solution of the einstein - matter field equations describing spherically symmetric shells satisfying an equation of state in closed form is discussed under general assumptions of physical reasonableness .
the solutions split into two classes : a class of astro - physically interesting " solutions describing ordinary " matter with positive density and pressure , and a class of phantom - like " solutions with positive density but negative active gravitational mass , which can also be of interest in several very strong fields " regimes .
known results on linear - barotropic equations of state are recovered as particular cases .
# 1#2 [ section ] [ teo]proposition [ teo]lemma [ teo]corollary [ teo]definition [ teo]remark [ teo]assumption [ teo]example |
let @xmath0 be a unital commutative associative algebra over a field @xmath5 with @xmath6 and @xmath1 be a @xmath5-lie algebra .
then the tensor product @xmath3 is a lie algebra with respect to the bracket @xmath7 : = aa ' \otimes [ x , x'].\ ] ] let @xmath2 be a vector space , considered as a trivial @xmath8-module .
the main point of the present paper is to give a description of the set @xmath4 of cohomology classes of @xmath2-valued @xmath9-cocycles on the lie algebra @xmath8 in terms of data associated to @xmath0 and @xmath1 which is as explicit as possible .
we consider @xmath2-valued @xmath9-cochains on @xmath8 as linear functions @xmath10 .
such a function is a _
@xmath9-cocycle _ if and only if it vanishes on the subspace @xmath11 of @xmath9-boundaries , which is the image of the linear map @xmath12 \wedge z + [ y , z ] \wedge x + [ z , x ] \wedge y.\ ] ] in view of the jacobi identity , @xmath11 is contained in the subspace @xmath13 of @xmath9-cycles , i.e. , the kernel of the linear map @xmath14.$ ] the quotient space @xmath15 is the _ second homology space of @xmath8_. a @xmath9-cocycle @xmath16 is a _
coboundary _ if it is of the form @xmath17)$ ] for some linear map @xmath18 .
we write @xmath19 for the set of @xmath9-coboundaries and @xmath20 for the set of @xmath9-cocycles .
this means that a coboundary vanishes on @xmath13 .
if , conversely , a @xmath9-cocycle vanishes on @xmath13 , then there exists a linear map @xmath21 \to \z$ ] with @xmath22 , and any linear extension @xmath23 of @xmath24 to all of @xmath8 yields @xmath25 .
this leads to the following description of the second @xmath2-valued cohomology group @xmath26 from this picture , it is clear that we obtain a good description of @xmath4 if we have an accessible description of the space @xmath13 and its subspace @xmath11 , hence of the quotient space @xmath27 .
our goal is a description of this space and the cocycles in terms of accessible data attached to the commutative algebra @xmath0 and the lie algebra @xmath1 . for a @xmath5-vector space @xmath28
, we identify the second exterior power @xmath29 , resp .
, the second symmetric power @xmath30 with the corresponding subspaces of @xmath31 .
accordingly , we put @xmath32 and obtain @xmath33 .
for the commutative algebra @xmath0 , we have a natural decomposition @xmath34 where @xmath35 is the kernel of the multiplication map @xmath36 .
the first step , carried out in section @xmath9 , is to show that by identifying @xmath0 with @xmath37 , we obtain a linear isomorphism @xmath38 restricting to a linear isomorphism @xmath39 now each alternating map @xmath10 is represented by three maps @xmath40 determined by @xmath41 in the sense of ( [ eq:1 ] ) . since two cocycles
define the same cohomology class if and only if they coincide on the subspace @xmath13 of @xmath42 , any cohomology class @xmath43 \in h^2(\g,\z)$ ] is represented by the triple @xmath44 , where @xmath45 .
conversely , three linear maps @xmath46 and @xmath47 as in ( [ eq:3 ] ) define a cocycle if and only if @xmath48 vanishes on @xmath11 .
the main result of the present paper is theorem [ theo2 ] which makes this condition more explicit as follows : \(a ) the alternating linear map @xmath49 defined by @xmath50 has values in the set @xmath51 of invariant symmetric bilinear maps and @xmath52 vanishes on @xmath53 , where @xmath54 and @xmath55 $ ] denotes the commutator algebra of @xmath1 .
\(b ) for the map @xmath56 defined by @xmath57 we have @xmath58,z ) \quad \mbox { for all } \quad a \in a , x , y , z \in \k,\ ] ] with the lie algebra differential @xmath59 .
\(c ) @xmath47 vanishes on @xmath60 .
note that these conditions imply that the two maps @xmath61 and @xmath47 are also cocycles , whereas @xmath52 and @xmath62 are cocycles if and only if @xmath52 vanishes on @xmath63 , which , in view of ( b ) , means that @xmath64 vanishes on @xmath65 , i.e. , @xmath66 has values in the space @xmath67 of @xmath2-valued @xmath9-cocycles on @xmath1 .
cocycles of the form @xmath61 , where @xmath52 and @xmath62 are not cocycles , are called _ coupled_. all coboundaries are of the form @xmath68 ( @xmath69 ) , so that the cohomology class of a coupled cocycle contains only coupled cocycles .
we show that @xmath8 possesses non - zero coupled @xmath9-cocycles if and only if the image of the universal derivation @xmath70 is non - trivial and @xmath71 possesses a symmetric invariant bilinear form @xmath72 for which the @xmath73-cocycle @xmath74,z)$ ] is a non - zero coboundary . the map @xmath75 is called the _ koszul map _
@xcite , 11 ; see also @xcite , p.113 )
. calling an invariant symmetric bilinear form @xmath76 _ exact _ if @xmath77 is a coboundary , this means that @xmath1 possesses exact invariant bilinear forms @xmath72 with @xmath77 non - zero .
note that this is not the case if @xmath1 is finite - dimensional semisimple , so that there are no coupled cocycles in this case .
our approach leads us to an exact sequence of the form @xmath78 which is the main result of section 4 . here
@xmath79 denotes the set of alternating bilinear forms on @xmath80 of the form @xmath81 , and for two pairs @xmath82 and @xmath83 of linear spaces with @xmath84 and @xmath85 we write @xmath86 so that we have @xmath87 , and we put @xmath88 and @xmath89 from the exact sequence , it follows that a crucial part of the description of @xmath90 lies in an understanding of the spaces @xmath91 and @xmath92 . in an appendix
, we show that the map @xmath93 , induced by the koszul map @xmath94 , is part of an exact sequence @xmath95 which implies that for the space @xmath96 of exact invariant forms we have @xmath97 in section [ sec5 ] , we give an example of a non - trivial coupled @xmath9-cocycle and in section [ sec6 ] we explain how our results can be used for the analysis of continuous cocycles if @xmath98 and @xmath0 and @xmath1 are locally convex spaces with continuous algebra structures .
then @xmath99 carries the structure of a locally convex lie algebra , and we are interested in the space @xmath100 of cohomology classes of continuous @xmath9-cocycles with values in a locally convex space @xmath2 modulo those coboundaries coming from continuous linear maps @xmath101 .
the main difficulty in applying the algebraic results in the topological context with an infinite dimensional lie algebra @xmath1 is the possible discontinuity of a linear map @xmath102 bounding an algebraically trivial @xmath9-cocycle . if @xmath71 is a finite dimensional semi - simple lie algebra and @xmath0 a topological algebra , then the continuous second cohomology space @xmath103 has been determined in @xcite as @xmath104 , where @xmath105 denotes the space of continuous @xmath5-valued cyclic @xmath106-cocycles on @xmath0 ( see @xcite for the algebraic case ) . as any exact form vanishes on a semi - simple lie algebra , there are no coupled cocycles in this case .
the main previous contributions to the investigations of @xmath27 for @xmath99 and arbitrary @xmath1 and @xmath0 are the articles by haddi @xcite and zusmanovich @xcite .
both offer a description of @xmath27 in terms of ( sub- or quotient ) spaces .
haddi @xcite uses the projection @xmath107 of @xmath108 to the homology of the quotient complex of @xmath71-coinvariants and computes kernel and cokernel of this map .
the cokernel of @xmath109 is isomorphic to @xmath110 $ ] , and the kernel is isomorphic to @xmath111 , @xmath112 is the kernel of the projection @xmath113 ( the subspace of _ essential homology _ ) , and @xmath114 is the subspace of @xmath108 generated by cycles of the form @xmath115 for @xmath116 or @xmath117 and @xmath118 , which lies in @xmath119 ( in our notation ) . furthermore he uses a non - canonical splitting to identify the homology of the coinvariants @xmath120 with @xmath121 , where @xmath122 is the image of @xmath123 in the space of @xmath71-coinvariants of symmetric @xmath9-tensors on @xmath71 .
he thus obtains an exact sequence @xmath124 \to 0.\ ] ] it is instructive to compare this sequence with our exact cohomology sequence described above .
zusmanovich @xcite uses as extra data a free presentation of @xmath71 and deduces one of @xmath8 .
he describes the subspace of essential homology @xmath125 by the hopf formula in terms of the presentation . in this way
, he identifies the different terms in the exact sequence given by the @xmath126-term exact sequence of the hochschild serre spectral sequence for the subalgebra @xmath127 ( using non - canonical splittings ) .
his description yields @xmath128 where @xmath129 is the space of @xmath71-coinvariants in @xmath130 , and @xmath131 is spanned by the elements @xmath132 for @xmath133 .
the main advantage of our approach is that is does not require any auxiliary data and provides a quite explicit description of cocycles representing the different types of cohomology classes . in particular
, this direct approach leads us to the interesting new class of coupled cocycles . in subsequent work
, we plan to use the methods developed in @xcite to study global central extensions of lie groups @xmath134 whose lie algebras are of the form @xmath99 defined by coupled lie algebra cocycles . for algebras of the type @xmath135 ,
i.e. , compactly supported smooth functions on a manifold @xmath136 , this has been carried out in @xcite and @xcite . *
thanks : * we are grateful to m. bordemann for a stimulating email exchange and for pointing out the relation to the exact sequence ( [ exseq0 ] ) , part of which is due to him .
we also thank the referee for an extremely valuable report and in particular for pointing out several references and some inaccuracies in a previous version . in the following ,
we write elements of @xmath137 simply as @xmath138 to simplify notation
. elements of @xmath0 are mostly denoted @xmath139 or @xmath140 and elements of @xmath71 are denoted @xmath141 or @xmath142 .
we write @xmath55 $ ] for the commutator algebra of @xmath1 and observe that @xmath143 is the commutator algebra of @xmath8 .
we also write @xmath144 , @xmath145 , @xmath146 , and @xmath147 for the spaces of lie algebra @xmath148-cochains , cocycles , coboundaries and cohomology classes with values in the trivial module @xmath5 .
we write @xmath149 for the space of @xmath2-valued symmetric bilinear maps on @xmath1 and put @xmath150 .
accordingly , we write @xmath151 for the set of @xmath2-valued alternating bilinear maps .
in this section , we review different constructions of the universal differential module @xmath152 .
the relationship between these constructions will play a crucial role in the following .
an important object attached to the algebra @xmath0 is its universal differential module @xmath152 .
this is an @xmath0-module with a derivation @xmath70 which is universal in the sense that for any other @xmath0-module @xmath136 and any derivation @xmath153 , there exists a unique module morphism @xmath154 of @xmath0-modules with @xmath155 . from its universal property , it is easy to derive that the universal differential module is unique up to isomorphism , but there are many realizations , looking at first sight quite differently
. let @xmath156 denote the multiplication of @xmath0 .
then @xmath157 is an algebra morphism , so that @xmath158 is an ideal of the commutative algebra @xmath159 . from the @xmath0-module structure on @xmath159 , given by @xmath160
, we thus derive an @xmath0-module structure on the quotient space @xmath161 , which also is a ( non - unital ) commutative algebra .
let @xmath162 $ ] denote the image of @xmath163 in @xmath161
. then @xmath164\ ] ] is a derivation and it is not hard to verify that @xmath165 has the universal property of @xmath166 ( cf .
@xcite , ch .
iii , 10.11 ) .
we obviously have the direct decomposition @xmath167 , where the projection onto the subspace @xmath168 is given by @xmath169 this implies that @xmath170 and thus @xmath171 another way to construct @xmath152 is by observing that each linear map @xmath153 leads to a linear map @xmath172 and that @xmath173 is a derivation if and only if @xmath174 which implies that @xmath175 contains the @xmath0-submodule @xmath176 of @xmath159 .
the quotient @xmath177 is called the _ first hochschild homology space of @xmath0_. from the preceding discussion
, it follows that the map @xmath178 \mapsto a d_a(b ) \end{aligned}\ ] ] is an isomorphism of @xmath0-modules because the map @xmath179 $ ] is a derivation with the universal property ( cf .
@xcite , prop .
the link between the description of @xmath152 as @xmath180 and @xmath161 is given by the commutative diagram @xmath181 with the isomorphism @xmath182 ) = a d(b ) = [ a \otimes b - ab \otimes \1]$ ] .
note that the commutativity of the diagram implies that @xmath183 let @xmath184 denote the image of the subspace @xmath185 under the quotient map @xmath186 . in view of @xmath187 , the image of the subspace of symmetric tensors , which we identify with @xmath188 , in @xmath152 coincides with @xmath189 , so that ( [ hhomega ] ) immediately shows that the map @xmath190 \mapsto [ a d_a(b)]\ ] ] induces a linear isomorphism .
it is well known that the _ first cyclic homology space _
@xmath191 = \lambda^2(a)/t(a)\ ] ] is of central importance for lie algebra @xmath9-cocycles on lie algebras of the form @xmath192 ( cf .
@xcite ) .
alternating bilinear maps @xmath193 for which the corresponding map @xmath194 vanishes on @xmath195 are called _
cyclic @xmath106-cocycles _ , which means that @xmath196 from the above , it follows that the space @xmath197 of @xmath2-valued cyclic @xmath106-cocycles can be identified with @xmath198 we define two trilinear maps @xmath199 and @xmath200 we also put @xmath201 . [ lem1.1 ]
the map @xmath202 is surjective and @xmath203 .
that @xmath204 is surjective follows from @xmath205 for the determination of the kernel of @xmath204 , we use the realization of @xmath152 as @xmath161 . in this case , @xmath206 $ ] , so that @xmath207 = [ a \otimes b - b \otimes a].\ ] ] therefore the kernel of @xmath204 is the intersection @xmath208 where we consider @xmath209 as the subspace of skew - symmetric tensors in @xmath159 . writing @xmath159 as @xmath210 , the commutativity of the multiplication of @xmath0 shows that @xmath211 since the flip involution is an algebra isomorphism of @xmath159 , we have @xmath212 this implies that @xmath213 and that this subspace coincides with the image of @xmath214 under the projection @xmath215 finally , this leads with ( [ eq:5 ] ) to @xmath216
in this section , we turn to the identification of the space @xmath11 of @xmath9-coboundaries in @xmath13 in terms of our threefold direct sum decomposition ( [ eq:2 ] ) . from the universal property of @xmath42
, we immediately obtain linear maps @xmath217 and @xmath218 we likewise have linear maps @xmath219 and @xmath220 satisfying @xmath221 in this sense , we have @xmath222 and the projections on the two summands are given by @xmath223 .
recall the kernel @xmath168 of the multiplication map @xmath224 .
the map @xmath225 is a section of the multiplication map @xmath157 , so that we obtain a direct sum decomposition @xmath226 ( cf .
( [ ja - dec ] ) ) . in view of this decomposition ,
we obtain a linear isomorphism @xmath227 where the projections @xmath228 on the three summands are given by @xmath229 @xmath230 the following lemma provides the decomposition of @xmath13 which is a central tool in the following .
[ lem2.1 ] the space @xmath13 is adapted to the direct sum decomposition of @xmath42 : @xmath231 since @xmath232 $ ] is symmetric in @xmath233 and alternating in @xmath234 , its kernel contains @xmath119 .
the formula for @xmath235 also shows immediately that @xmath236 , so that it remains to observe that @xmath237 because @xmath238 = a b_\k(x \wedge y).$ ] in the following , we write @xmath239 for congruence of elements of @xmath42 modulo @xmath11 . [ lem2.2 ] for @xmath240 and @xmath241 , we have @xmath242 ) \equiv - p_2(\partial(ax\wedge by \wedge cz ) ) \mod b_2(\g).\ ] ] in particular , @xmath243 from @xmath244 \wedge cz + bc[y , z ] \wedge ax + ac [ z , x ] \wedge by \\ \partial(cx \wedge ay \wedge bz ) & = & ac[x , y ] \wedge bz + ab[y , z ] \wedge cx + bc [ z , x ] \wedge ay \\
\wedge ac y \wedge bz ) & = & ac[x , y ] \wedge bz + abc[y , z ] \wedge x + b [ z , x ] \wedge acy \\
\wedge y \wedge abz ) & = & c[x , y ] \wedge abz + ab[y , z ] \wedge cx + abc [ z , x ] \wedge y \\
\partial(bcx \wedge ay \wedge z ) & = & abc[x , y ] \wedge z + a[y , z ] \wedge bcx + bc [ z , x ] \wedge ay \\
\wedge y \wedge z ) & = & abc[x , y ] \wedge z + [ y , z ] \wedge abcx + abc [ z , x ] \wedge y , \\\end{aligned}\ ] ] we derive @xmath245 \wedge cz + bc[y , z ] \wedge ax + ac [ z , x ] \wedge by - b[z , x ] \wedge ac y - c[x , y ] \wedge abz - a [ y , z ] \wedge bc x \\ & & - abc[y , z ] \wedge
x + [ y , z ] \wedge abcx \\ & = & 2 ab \wedge c \otimes [ x , y ] \vee z + 2 bc \wedge a \otimes [ y , z ] \vee x + 2 ac \wedge b \otimes [ z , x ] \vee y - 2 abc \wedge 1 \otimes x \vee [ y , z ] \\ & = & 2 p_1(\partial(ax \wedge by \wedge cz ) ) - 2 p_1(abc x \wedge [ y , z ] ) .
\end{aligned}\ ] ] this proves the first congruence .
note that for @xmath246 and @xmath241 we have @xmath247 \wedge z + [ y , z ] \wedge ax + a [ z , x ] \wedge y,\ ] ] which implies that @xmath248 \wedge z + a[z , x ] \wedge y \equiv ax \wedge [ y , z ] \mod b_2(\g ) .
\end{aligned}\ ] ] summing over all cyclic permutations of @xmath249 leads to @xmath250 \wedge z \equiv \sum_\cyc ax \wedge [ y , z ] \mod
b_2(\g ) . \end{aligned}\ ] ] from the relation ( [ rel1 ] ) , we get @xmath251 ) = abc[y , z ] \wedge x + abc x \wedge [ y , z ] \equiv \sum_\cyc abc[y , z ] \wedge
x = \sum_\cyc abc[x , y ] \wedge z.\ ] ] in view of @xmath252\wedge cz + bc[y , z]\wedge ax + ca[z , x]\wedge by ) \notag \\ & = & abc\otimes([x , y]\wedge z + [ y , z]\wedge x + [ z , x]\wedge y ) \notag \\ & = & abc \otimes \partial(x \wedge y \wedge z ) , \label{rel3}\end{aligned}\ ] ] relation ( [ rel2 ] ) yields @xmath253 \wedge z - \sum_\cyc abc z \wedge [ x , y ] \equiv \sum_\cyc abc[x , y ] \wedge z - 2 \sum_\cyc abc[x , y ] \wedge z \\ & = & -\sum_\cyc abc[x , y ] \wedge z \equiv - 2 p_1(\partial(ax \wedge by \wedge cz ) ) . \end{aligned}\ ] ] in view of the preceding lemma , the projection @xmath254 of @xmath42 onto the subspace @xmath255 preserves @xmath11 .
this also implies that @xmath256 preserves @xmath11 , and we derive that @xmath257 the following lemma provides refined information .
[ lem2.3 ] ( 1 ) : : @xmath258 and @xmath259 . ( 2 ) : : @xmath260 .
( 3 ) : : @xmath261 .
\(2 ) follows immediately from formula ( [ rel3 ] ) .
\(1 ) recall the identifications @xmath262 and @xmath263 .
that @xmath264 is contained in @xmath11 follows immediately from @xmath265 \wedge z + b[y , z ] \wedge ax + a[z , x ] \wedge by - ab[x , y ] \wedge z - a[y , z ] \wedge bx - b[z , x ] \wedge ay \\ & = & b[y , z ] \wedge ax + a[z , x ] \wedge by - a[y , z ] \wedge bx - b[z , x ] \wedge ay \\ & = & 2b \wedge a \otimes [ y , z ] \vee x + 2 a \wedge b \otimes [ z , x ] \vee y \\ & = & 2 a \wedge b \otimes ( [ z , x ] \vee y - [ y , z ] \vee x)\\ & = & 2 a \wedge b \otimes z.(x \vee y).\end{aligned}\ ] ] therefore the description of @xmath266 follows from @xmath267\wedge cz + bc[y , z]\wedge ax + ca[z , x]\wedge by ) \\ & = & ab\wedge c\otimes[x , y]\vee z + bc\wedge a\otimes[y , z]\vee x + ca\wedge b\otimes[z , x]\vee y \\ & \equiv & ( ab\wedge c + bc\wedge a + ca\wedge b)\otimes[x , y]\vee z\quad \mod \lambda^2(a)\otimes\k.s^2(\k ) \\ & = & t(a , b , c ) \otimes[x , y]\vee z. \end{aligned}\ ] ] in ( [ rel3 ] ) , we have seen that @xmath268 and this implies that @xmath269 which leads to @xmath270 in view of @xmath271 and lemma [ lem2.2 ] , the following element is contained in @xmath11 : @xmath272\vee z + \lambda^2(a)\otimes\k.s^2(\k ) \\ & \subeq & t_0(a , b , c)\otimes[x , y]\vee z + b_2(\g),\end{aligned}\ ] ] and now lemma [ lem2.2 ] implies that @xmath273 .
\(3 ) first we note that @xmath274 \wedge z + b[y , z ] \wedge ax + a[z , x ] \wedge by ) \notag \\ & = & ( ab \vee \1 - ab \vee \1 ) \otimes [ x , y ] \wedge z + ( b \vee a - ab \vee
\1 ) \otimes [ y , z ] \wedge x + ( a \vee b - ab \vee \1 ) \otimes [ z , x]\wedge y\notag \\ & = & ( a \vee b - ab \vee \1 ) \otimes ( [ y , z ] \wedge x + [ z , x]\wedge y ) .
\label{eq : sharp}\end{aligned}\ ] ] since @xmath275 preserves @xmath11 ( lemma [ lem2.2 ] ) , this expression lies in @xmath11 . using the same formula for all cyclic permutations of @xmath276 and adding all three terms
, we see that @xmath277 \wedge z \in b_2(\g).\ ] ] this also implies that @xmath278 \wedge z = ( a \vee b - ab \vee \1 ) \otimes \big(\sum_\cyc [ x , y ] \wedge z - ( [ y , z ] \wedge x + [ z , x]\wedge y)\big ) \in b_2(\g).\ ] ] next we note that @xmath279 is spanned by elements of the form @xmath280 because @xmath281 is the projection of @xmath188 onto @xmath279 with kernel @xmath282
. therefore @xmath11 contains @xmath283 . on the other hand , ( [ eq : sharp ] ) shows that @xmath284 is clearly contained in @xmath283 .
[ theo1 ] with the linear map @xmath285 \vee z ) + a \otimes \partial(x \wedge y \wedge z)\ ] ] we get the following description of @xmath11 : @xmath286 the description of the position of @xmath11 given in lemma [ lem2.3 ] is already quite detailed .
it shows in particular that @xmath287 and that @xmath288 contains @xmath264 . we know from the proof of lemma [ lem2.3](1 ) that , modulo the subspace @xmath289
, we have @xmath290 \vee z + abc \otimes \partial(x \wedge y \wedge z)\\ & = & t_0(a , b , c ) \otimes [ x , y ] \vee z + abc \wedge \1 \otimes [ x , y ] \vee z + abc \otimes \partial(x \wedge y \wedge z ) \\ & = & t_0(a , b , c ) \otimes [ x , y ] \vee z + f(abc \otimes x \otimes y \otimes z ) \subeq t_0(a ) \otimes \k \vee \k ' + \im(f ) .
\end{aligned}\ ] ] since @xmath291 by lemma [ lem2.3 ] , we also obtain the converse inclusion @xmath292 now the theorem follows .
as explained in the introduction , elements of @xmath4 can be identified with linear maps @xmath293 , vanishing on the subspace @xmath11 .
we further write @xmath9-cocycles as @xmath294 , according to the decomposition in lemma [ lem2.1 ] , where @xmath295 here @xmath52 corresponds to an alternating bilinear map @xmath49 , @xmath62 to a linear map @xmath296 and @xmath47 to a symmetric bilinear map @xmath297 the condition , that three such maps @xmath298 combine to a @xmath9-cocycle @xmath299 is that @xmath16 vanishes on @xmath11 . to make this condition more explicit , we define the _ koszul map _ @xmath300,z).\ ] ] that @xmath77 is alternating follows from @xmath301,y ) = \kappa(y,[x , z ] ) = \kappa([y , x],z ) = - \gamma(\kappa)(x , y , z)\ ] ] and the fact that the symmetric group @xmath302 is generated by the transpositions @xmath303 and @xmath304 . that the image of @xmath94 consists of @xmath73-cocycles is well known ( @xcite , 11 ; @xcite , p.113 ) . recall that for each @xmath1-module @xmath305 , the lie algebra differential @xmath306 is given by @xmath307 , x_0 , \ldots , \hat x_i , \ldots , \hat x_j , \ldots , x_p ) , \end{aligned}\ ] ] where @xmath308 indicates omission of @xmath309 . for the following theorem
, we observe that the lie algebra differential @xmath59 factors through the surjective map @xmath310 whose kernel are the @xmath9-coboundaries .
( description of cocycles)[theo2 ] the function @xmath294 as above is a @xmath9-cocycle if and only if the following conditions are satisfied : ( a ) : : @xmath311 .
( b ) : : @xmath312 vanishes on @xmath313 .
( c ) : : @xmath314 for each @xmath246 .
( d ) : : @xmath315 vanishes on @xmath313 .
the linear map @xmath16 is a @xmath9-cocycle if and only if it vanishes on @xmath11 . in view of theorem [ theo1 ]
, @xmath11 is the sum of four subspaces , so that we get four conditions .
condition ( a ) means that @xmath16 vanishes on @xmath264 , and condition ( b ) that it vanishes on the subspace @xmath316 .
that @xmath16 vanishes on the image of @xmath317 , means that @xmath318,z ) = - \tilde f_2(a)(\partial(x \wedge y \wedge z ) ) = ( d_\k\tilde f_2(a))(x , y , z)\ ] ] for @xmath246 and @xmath241 , which is ( c ) .
finally , ( d ) means that @xmath16 vanishes on @xmath319 .
@xmath294 is a cocycle if and only if @xmath320 and @xmath47 are cocycles .
[ cor2theo2 ] a function of one of the three types @xmath321 , @xmath322 , is a @xmath9-cocycle if and only if the following conditions are satisfied : ( @xmath323 ) : : @xmath311 and the induced map @xmath324 is a cyclic @xmath106-cocycle .
( @xmath325 ) : : @xmath326 .
( @xmath327 ) : : @xmath315 vanishes on @xmath313 . that @xmath321 is a @xmath9-cocycle is equivalent to @xmath16 vanishing on @xmath328 , so that lemma [ lem2.3 ] leads to the stated characterizations .
[ rem3.4 ] a special class of cocycles are those of the form @xmath329 , vanishing on @xmath330 .
the cocycles of the form @xmath331 also vanish on the commutator algebra , and the sums of these two types exhaust the image of the injective pull - back map @xmath332 where @xmath333 denotes the set of all alternating maps vanishing on @xmath334 .
[ cortheo2 ] for each cocycle @xmath294 , there exists a decomposition @xmath335 , where @xmath336 conditions ( a ) and
( b ) in theorem [ theo2 ] only refer to the restriction @xmath337 of @xmath52 to the subspace @xmath338 of @xmath119 .
this has the following interesting consequence .
we have a short exact sequence @xmath339 where the surjectivity of the map @xmath340 follows from the fact that any symmetric bilinear extension of an element of @xmath341 is invariant .
any splitting of this sequence extends @xmath342 to an alternating bilinear map @xmath343 with @xmath344 and such that @xmath345
. then @xmath346 so that @xmath347 also is a cocycle by theorem 3.1 .
we conclude that @xmath348 is a cocycle vanishing on @xmath330 .
this proves the assertion .
( description of coboundaries ) [ prop - cobound ] a cocycle @xmath294 is a coboundary if and only if @xmath69 and there exists a linear map @xmath349 with @xmath350 , i.e. , @xmath351 that @xmath16 is a coboundary means that it vanishes on @xmath13 . according to lemma [ lem2.1 ] , this implies that @xmath69 .
since the bracket map @xmath352 is alternating in @xmath1 and symmetric in @xmath0 , all coboundaries are of the form @xmath68 .
a _ coupled cocycle _ is a cocycle of the form @xmath320 for which @xmath52 is not a cocycle .
the following theorem characterizes the pairs @xmath353 for which @xmath192 possesses coupled cocycles . in section [ sec5 ] below , we shall also give a concrete example of a lie algebra @xmath1 satisfying this condition . [ theo3 ] the lie algebra @xmath99 possesses coupled cocycles if and only if @xmath354 and @xmath1 possesses a symmetric invariant bilinear form @xmath72 for which @xmath355 is a non - zero coboundary . if this is not the case , then each cocycle @xmath356 is a sum @xmath357 of four cocycles , where ( a ) : : @xmath358 vanishes on @xmath359 .
( b ) : : @xmath360 is a cyclic @xmath106-cocycle .
( c ) : : @xmath326 .
( d ) : : @xmath47 vanishes on @xmath330 .
first let @xmath361 be a coupled cocycle on @xmath8 .
then we have @xmath362 . composing with a suitable linear functional @xmath363 with @xmath364 we may w.l.o.g .
assume that @xmath365 .
then there exists an @xmath246 with @xmath366 now @xmath367 is an invariant symmetric bilinear form for which @xmath77 is exact and non - zero .
then @xmath368 ( theorem [ theo2 ] ) , so that @xmath369 in @xmath152 ( lemma [ lem1.1 ] ) .
if , conversely , @xmath354 and @xmath72 is an invariant symmetric bilinear form on @xmath1 for which @xmath77 is a non - zero coboundary , then we pick @xmath370 with @xmath371 .
we now define linear maps @xmath372 and @xmath373 we claim that the corresponding map @xmath361 is a @xmath9-cocycle by verifying the conditions in theorem [ theo2 ] .
condition ( a ) is obviously satisfied , and ( b ) follows from @xmath374 ( lemma [ lem1.1 ] ) .
further @xmath375 , and ( c ) follows from @xmath376 that @xmath52 is not a cocycle , i.e. , that @xmath16 is coupled , means that @xmath377 , which is equivalent to @xmath378 and @xmath379 this completes the proof of the first part of the theorem . for the second part , we assume that either @xmath380 vanishes , which means that @xmath381 , or that for each exact invariant symmetric bilinear form @xmath72 on @xmath1 we have @xmath382 .
then for each cocycle @xmath383 as in corollary [ cortheo2 ] , either @xmath384 vanishes on @xmath195 ( if @xmath189 vanishes ) or @xmath326 ( if for all exact forms on @xmath1 the @xmath73-cocycle @xmath77 vanishes ) .
both conditions imply that @xmath385 and @xmath62 are cocycles .
hence the assertion follows from corollary [ cor2theo2 ] .
if @xmath386 , then @xmath99 has no coupled cocycles . from the exact sequence in proposition [ transfer - seq ] below , it follows that the koszul map @xmath387\ ] ] is injective , and this implies that each exact invariant form vanishes .
the following proposition describes the universal cocycle for @xmath8 in terms of our threefold direct sum decomposition .
( a universal cocycle ) let @xmath388 denote a linear projection onto @xmath389 .
then the linear map @xmath390 maps @xmath11 into itself , hence induces a @xmath9-cocycle @xmath391 it is universal in the sense that for each space @xmath2 the map @xmath392 is a linear bijection . that @xmath393 is a linear projection onto @xmath13 follows from lemma [ lem2.1 ] .
the remainder follows from the fact that @xmath394 \mapsto f\res_{z_2(\g)}$ ] is injective onto the set of all maps vanishing on @xmath11 .
in this section , we use the results of the previous section to give a quite explicit description of the space @xmath395 in terms of data associated directly to @xmath8 and @xmath0 .
[ lem4.1 ] associating with each linear map @xmath396 the corresponding cocycle @xmath397 , we obtain , together with the natural pull - back map @xmath398 , an injection @xmath399 whose image consists of all classes of cocycles of the form @xmath400 .
the image of the pull - back map @xmath401 consists of those cohomology classes represented by cocycles vanishing on @xmath330 , which are the cocycles of the form @xmath402 . since the space of these cocycles intersects @xmath403
trivially , the space @xmath79 injects into @xmath90 ( remark [ rem3.4 ] and prop .
[ prop - cobound ] ) .
next we recall that the cocycles of the form @xmath404 correspond to linear maps @xmath396 ( which means that @xmath62 vanishes on @xmath405 ) , and that such a map is a coboundary if and
only if @xmath406 , because this implies the existence of a linear map @xmath407 with @xmath408 for all @xmath246 . the latter condition means that @xmath62 vanishes on @xmath409 , so that the cohomology classes correspond to elements in @xmath410 given a cocycle @xmath294 in @xmath411 , we obtain the map @xmath412 whose kernel contains @xmath413 , so that it induces a linear map @xmath414 mapping the subspace @xmath415 into the subspace @xmath416 ( theorem [ theo2 ] ) . in view of @xmath417 the range of each map @xmath418 lies in the subspace @xmath88 and @xmath419
we thus obtain a map @xmath420 \mapsto \gamma \circ \tilde f_1,\ ] ] where for pairs @xmath82 and @xmath83 of linear spaces with @xmath84 and @xmath85 we write @xmath421 [ theo4 ] the sequence @xmath422 is exact .
we have already seen in lemma [ lem4.1 ] that @xmath423 is injective .
the kernel of @xmath424 consists of all cocycles @xmath294 for which @xmath425 .
this is equivalent to @xmath426 vanishing on @xmath427 , which means that @xmath52 vanishes on @xmath330 , i.e. , @xmath428 .
we conclude that @xmath429 . to see that @xmath424 is surjective , let @xmath430 and observe that there exists a linear map @xmath431 and a linear map @xmath432 with @xmath433 for @xmath434 we then have @xmath435 so that the corresponding maps @xmath52 and @xmath62 sum up to a @xmath9-cocycle @xmath436 satisfying @xmath437 ) = \alpha$ ] . the quotient @xmath438 can be identified with the image of the map @xmath439\ ] ] discussed in the appendix below . from the exactness of the sequence in proposition [ transfer - seq ] , it follows that the space @xmath440 of exact invariant bilinear forms satisfies @xmath441 we also note that for a _
quadratic lie algebra _ , i.e. , a finite - dimensional lie algebra @xmath1 with an invariant non - degenerate symmetric bilinear form @xmath442 , the space @xmath443 of _ outer derivations _ satisfies @xmath444 and that the subspace @xmath445 consists of those classes @xmath446 $ ] of derivations @xmath173 which are skew - symmetric with respect to @xmath442 .
we further have @xmath447 , so that @xmath448 to obtain an explicit description of @xmath90 , it is therefore necessary to have a good description of the space @xmath449 of invariant quadratic forms on @xmath1 and its subspace of exact forms .
let @xmath1 be a finite - dimensional @xmath5-lie algebra .
we consider the space @xmath450 of invariant symmetric bilinear forms on @xmath1 .
let @xmath451 denote the common radical of all invariant symmetric bilinear forms on @xmath1 .
fix an element @xmath452 of maximal rank .
then @xmath453 , but is there some @xmath72 for which we have equality ? in the following remark , we collect some information that is useful to determine the space @xmath91 .
suppose that @xmath454 is a quadratic lie algebra , i.e. , @xmath442 is a non - degenerate invariant symmetric bilinear form on @xmath1
. then there exists for each invariant symmetric bilinear form @xmath455 a uniquely determined endomorphism @xmath456 with @xmath457 now the invariance of @xmath72 implies that @xmath458 is contained in the _ centroid _
@xmath459 = 0\}.\ ] ] the centroid of @xmath1 is an associative subalgebra of @xmath460 on which transposition @xmath461 with respect to @xmath442 induces a linear anti - automorphism , satisfying @xmath462 it follows in particular that for @xmath463 the invariant bilinear form @xmath464 is symmetric if and only if @xmath465 .
this leads to a linear bijection @xmath466 for @xmath467 , the invariant form @xmath468 is alternating , which implies that @xmath468 vanishes on @xmath313 , and this implies that @xmath469 conversely , any @xmath470 with @xmath471 and @xmath472 satisfies @xmath473 for all @xmath474 , hence is contained in the centroid .
we put @xmath475 and observe that @xmath476 is an ideal of the associative algebra @xmath477 because @xmath478 is the kernel of the restriction homomorphism @xmath479 .
if @xmath480 , then @xmath481,\k ) = \kappa_0([\k,\k],a.\k ) \subeq \kappa_0(\k ' , \z(\k ) ) = \{0\},\ ] ] so that @xmath482 .
hence the ideal @xmath483 is invariant under transposition .
we have already seen that @xmath483 contains all skew - symmetric elements of @xmath477 , so that the involution induced on the quotient algebra @xmath484 is trivial , which implies that this algebra is commutative .
we thus have
@xmath485 therefore @xmath486 carries the structure of an associative commutative algebra . in @xcite ,
th . 2.3 ,
medina and revoy describe the structure of the associative algebra @xmath487 for a lie algebra @xmath71 whose center @xmath488 is contained in @xmath489 : the algebra @xmath487 has a decomposition with respect to orthogonal indecomposable idempotents @xmath490 with @xmath491 , so that @xmath71 is the direct product of the ideals @xmath492 .
moreover , the algebra @xmath493 is a local ring , and we have @xmath494 as linear spaces , and @xmath495 if , in addition , @xmath1 carries a non - degenerate quadratic from @xmath442 , then th .
_ implies that the decomposition of @xmath71 as a direct sum of ideals @xmath496 is orthogonal and the idempotents @xmath497 are symmetric with respect to @xmath442 .
we conclude in particular that @xmath498
in this section we describe some lie algebras @xmath1 on which we have invariant bilinear forms @xmath72 for which @xmath77 is a non - zero coboundary , so that @xmath99 has coupled cocycles whenever @xmath499 .
let @xmath500 be the @xmath73-dimensional heisenberg algebra @xmath501 with generators @xmath116 , @xmath502 and @xmath503 and the only non - trivial relation @xmath504=c$ ] . then pass to the extension @xmath505 of @xmath501 by a derivation @xmath173 like for affine kac
moody algebras .
explicitly , we take @xmath506 , @xmath507 and @xmath508 ( cf .
@xcite , p.98 , ex .
@xmath509 ) .
we write @xmath510 for the element of @xmath1 corresponding to @xmath173 .
the lie algebra @xmath1 is @xmath511-dimensional , and has an invariant bilinear symmetric form @xmath72 , as any lie algebra with symmetrizable cartan matrix ( cf .
@xcite , prop . 4 , p. 362 ) .
we call @xmath1 the _ split oscillator algebra over @xmath5_. let us compute the dimensions of the spaces of cochains , cocycles and cohomology spaces : [ cols="^,^,^,^,^,^",options="header " , ] in the preceding table , the dimension of the cohomology spaces is computed as follows : @xmath512 by definition . as @xmath513\,=\,\k d$ ] , @xmath514 . by unimodularity , @xmath1 satisfies poincar duality ( @xcite , p. 27 ) , so that the dimensions in degree @xmath73 and @xmath511 follow .
but the euler characteristic of a finite dimensional lie algebra vanishes @xcite , which implies that @xmath515 .
the dimensions of the boundary spaces are clear in degree @xmath516 and @xmath106 . in degree
@xmath9 , there remain @xmath73 dimensions as the difference of @xmath517 and @xmath518 . in the same way
, we get the dimensions of @xmath519 for @xmath520 .
finally , @xmath521 is the sum of @xmath522 and @xmath523 .
observe that @xmath524=\fh$ ] and @xmath525=\k c$ ] , so that @xmath1 is solvable , but @xmath526=\fh$ ] , so that @xmath1 is not nilpotent .
we claim that each invariant bilinear form @xmath72 is exact , which gives rise to coupled cocycles ( in the sense of section @xmath73 ) : if @xmath527 , then the fact that @xmath1 is unimodular implies that all @xmath73-cochains @xmath528 , @xmath529 , are @xmath73-cocycles .
if @xmath530 = \h$ ] , then @xmath528 is exact , so that @xmath531 yields a basis of the one - dimensional space @xmath532 . since @xmath533 and for each invariant symmetric bilinear form @xmath72 we have @xmath534,c ) = \kappa(x,[y , c])=0 $ ] , we see that @xmath535 .
hence each invariant symmetric bilinear form is exact .
we now turn to the space @xmath449 : any invariant symmetric bilinear form @xmath72 satisfies @xmath536,x)\,=\,-\kappa([y , x],x)\,=\,-\kappa(y,[x , x])\,=\,0.\ ] ] @xmath537,y)\,=\,\kappa(x,[y , y])\,=\,0.\ ] ] @xmath538,c)\,=\,\kappa(x,[y , c])\,=\,0.\ ] ] @xmath539)\,=\,\kappa([d , d],x)\,=\,0.\ ] ] @xmath540)\,=\,-\kappa([d , d],y)\,=\,0.\ ] ] @xmath541)\,=\,\kappa([d , x],y)\,=\,\kappa(x , y).\ ] ] @xmath542)\,=\,-\kappa(x,[x , d])\,=\,-\kappa([x , x],d)\,=\,0.\ ] ] @xmath543)\,=\,\kappa(y,[y , d])\,=\,\kappa([y , y],d)\,=\,0.\ ] ] we immediately conclude that the space of invariant symmetric bilinear forms is at most @xmath9-dimensional and that each such form @xmath72 is determined by @xmath544 and @xmath545 ( note that @xmath546 is not a commutator ) .
let us denote by @xmath547 the ( invariant symmetric bilinear form ) with @xmath548 and @xmath549 and @xmath550 the invariant symmetric bilinear form with @xmath551 and @xmath552 .
then @xmath550 coincides with the invariant form @xmath72 introduced above and @xmath553 form a basis of @xmath449 .
combining with the observation in the preceding remark and section 4 , we get @xmath554 for the reduced centroid , we thus get @xmath555 ) = \k[\kappa_2].\ ] ] we further get @xmath556 . for any algebra @xmath0 , and @xmath99 ,
the exact sequence in theorem 4.2 now turns into a sequence of the form @xmath557 therefore the essential part of @xmath90 is isomorphic to the dual space of @xmath152 . from the construction in the proof of theorem 4.2
, it follows that the coupled cocycles correspond to the elements of @xmath558 not vanishing on the subspace @xmath189 . for a lie algebra @xmath8 its cotangent bundle
@xmath559 is a lie algebra with the bracket @xmath560 : = ( x.f ' - x'.f,[x , x ' ] ) , \quad \hbox { where } \quad x.f = \ad^*x.f = -f \circ \ad x.\ ] ] a slight generalization is obtained as follows .
suppose that @xmath561 is a lie algebra @xmath9-cocycle , i.e. , @xmath562,z ) = 0.\ ] ] then we have another lie algebra structure on @xmath563 given by @xmath560 : = ( x.f ' - x'.f + \gamma(x , x'),[x , x']).\ ] ] we write @xmath564 for the corresponding lie algebra , a so - called _ twisted magnetic extension of @xmath8_. the symmetric bilinear form given by @xmath565 satisfies @xmath566 ) , ( f'',x '' ) ) \cr & = & f''([x , x ' ] ) + f'([x'',x ] ) + f([x',x '' ] ) + \gamma(x , x')(x '' ) \cr & = & \gamma(x , x')(x '' ) + \sum_\cyc f([x',x '' ] ) .
\end{aligned}\ ] ] this implies that @xmath72 is invariant if and only if @xmath567 is alternating , hence an element of @xmath568 ( cf .
@xcite , lemma 3.1 ) .
let us assume that this is the case and note that @xmath77 vanishes only if @xmath569 and @xmath8 is abelian .
for the alternating bilinear form @xmath570 we then have @xmath571 , ( f'',x '' ) ) & = & ( x.f ' - x'.f + \gamma(x , x'))(x '' ) - f''([x , x ' ] ) \cr & = & \gamma(x , x')(x '' ) + f'([x'',x ] ) + f([x',x '' ] ) - f''([x , x ' ] ) , \end{aligned}\ ] ] so that @xmath572 ) + f([x',x '' ] ) - f''([x , x'])\big ) - \sum_\cyc \gamma(x , x')(x '' ) \cr & = & - 3 \gamma(x , x')(x '' ) - \sum_\cyc f([x',x '' ] ) .
\end{aligned}\ ] ] let @xmath573 denote the canonical projection
. then the preceding calculation shows that @xmath77 is a coboundary if and only if @xmath574
\g)$ ] vanishes .
this is in particular the case for @xmath575 . in @xcite , pelc introduces a family of lie algebras @xmath576 , where @xmath576 is an @xmath577-dimensional lie algebra with basis @xmath578 and commutator relations
@xmath579 = \left\ { \begin{array}{cl } \hat{i - j } \cdot t_{i+j } & \mbox{for $ i + j \leq n$ } \\ 0 & \mbox { otherwise . }
\end{array } \right.\ ] ] here @xmath580 is chosen such that @xmath581 .
then @xmath576 is a solvable lie algebra with commutator algebra @xmath582 .
let us assume that @xmath583 for some @xmath584 and that @xmath585 .
then @xmath586 carries a non - degenerate invariant symmetric bilinear form defined by @xmath587 ( cf .
we claim that @xmath72 is exact .
choose @xmath588 in such a way that @xmath589 , where @xmath590 and @xmath591 .
then @xmath592 and @xmath593 , and we define a @xmath9-cochain @xmath594 by @xmath595 note that we need @xmath596 to see that this is well - defined .
now @xmath597 , t_k)$ ] vanishes if @xmath598 , and for @xmath599 we get @xmath600 , t_k ) = \hat{i - j } \cdot a_{i+j}.\ ] ] with @xmath601 this leads to @xmath602 , t_k).\ ] ] therefore @xmath72 is exact . note that for @xmath603 we thus obtain the split oscillator algebra .
\(a ) if @xmath604 is the split oscillator algebra from section 5.1 , then @xmath605 for the @xmath9-dimensional non - abelian subalgebra @xmath606 and th . 3.2 in @xcite implies that @xmath607 is isometrically isomorphic to @xmath608 , which provides another argument for the exactness of @xmath550 .
\(b ) for @xmath583 and @xmath609 , we observe that the ideal @xmath610 is isotropic and abelian , so that @xcite , thm .
3.2 implies that @xmath611 for @xmath612 .
therefore pelc s algebras provide other examples of exact twisted cotangent bundles . using the canonical basis @xmath613 of @xmath1 to define a section @xmath614
, we obtain @xmath615 in particular , @xmath616 is non - zero .
in this section , we explain how the algebraic results from the preceding sections can be used in the topological setting .
actually these applications were our original motivation to study the work of haddi and zusmanovich .
we now assume that @xmath617 or @xmath618 .
let @xmath0 be a unital commutative locally convex associative @xmath5-algebra and @xmath1 a locally convex @xmath5-lie algebra .
we endow @xmath99 with the projective tensor product topology , turning it into a locally convex space with the universal property that for each locally convex space @xmath2 a bilinear map @xmath619 is continuous if and only if the corresponding linear map @xmath620 is continuous
. then the lie bracket on @xmath8 is continuous because the quadrilinear map @xmath621\ ] ] is continuous and the continuous quadrilinear maps correspond to the continuous linear maps on @xmath622 . in the topological context
, we consider for a locally convex space @xmath2 the space @xmath623 of continuous cocycles and the subspace @xmath624 of all coboundaries of the form @xmath625 , where @xmath18 is a continuous linear map . in the topological context ,
the relation between the space @xmath626 and the space of all linear maps from @xmath627 is more complicated than in the algebraic setup ( @xcite ) . to define the topological version of @xmath27
, we have to use the closure of @xmath11 to obtain a hausdorff topology on the quotient space .
we always have a natural map @xmath628 but in general there is no reason for this map to be injective or surjective .
therefore the homology space is much less interesting in the topological setting , and it often is easier to work directly with cocycles and coboundaries which is made possible by our results in section iii . the flip involution on @xmath629 , endowed with the projective tensor product topology , is continuous , so that the kernel of the quotient map @xmath630 is closed , which leads to a locally convex topology on @xmath42 .
further the bracket map @xmath352 is continuous because it is induced from the continuous bracket map , which shows that its kernel @xmath13 is closed .
one easily verifies that the maps @xmath223 and @xmath631 from section 2 are continuous , and likewise that the maps @xmath632 are continuous .
therefore lemma [ lem2.1 ] yields a topological decomposition of the closed subspace @xmath13 of @xmath42 : @xmath633 this implies that any continuous cocycle @xmath634 defines three continuous maps @xmath635 conversely , three such continuous linear maps combine to a continuous @xmath9-cocycle of @xmath8 if and only if they satisfy the conditions from theorem [ theo2 ] .
if a continuous cocycle @xmath636 is contained in @xmath624 , then it vanishes on @xmath13 , which implies @xmath69 and that @xmath62 is a continuous coboundary , i.e. , there exists a continuous linear map @xmath18 with @xmath637 ) \quad \mbox { for all } \quad a , b\in a , x , y \in \k.\ ] ] clearly , this implies that @xmath638 . if , conversely , @xmath638 , then there exists a linear map @xmath639 with @xmath640 for all @xmath246 , but it is not clear whether the corresponding map @xmath641 will be continuous . therefore the exactness condition is quite subtle .
if @xmath1 is finite - dimensional , then the situation simplifies significantly .
then @xmath642 and if @xmath643 stands for `` continuous linear maps '' , then @xmath644 so that we may consider @xmath62 as a @xmath9-cocycle in @xmath645 .
if this map vanishes on @xmath65 , then there exists a linear map @xmath646 with @xmath647)(a ) , \quad x , y \in \k , a \in a.\ ] ] then the map @xmath648 is continuous and satisfies @xmath649 .
we thus get @xmath650 we collect the previous remarks in the following theorem which is analoguous to theorem [ theo4 ] .
it determines the structure of the second continuous cohomology space for current algebras .
let us denote by @xmath651 the locally convex module of khler differentials for the locally convex commutative associative algebra @xmath0 .
let @xmath1 be a finite dimensional lie algebra over @xmath5 .
then the sequence @xmath652 is exact .
first we note that the short exact sequence @xmath653 of finite - dimensional vector spaces splits .
since @xmath143 is closed in @xmath8 , it follows that the short exact sequence @xmath654 also splits topologically .
as we have observed above , theorem [ theo3 ] and its corollaries remain true in the topological setting . for corollary [ cortheo2 ]
, we use the topological splitting of @xmath655 in @xmath1 .
we have also seen above that the corresponding description of the coboundaries remains valid because @xmath1 is finite dimensional .
further , the topological splitting of @xmath656 implies that lemma [ lem4.1 ] remains true .
this implies the injectivity of @xmath423 . that @xmath657 is shown as in the proof of theorem [ theo4 ] .
finally , the surjectivity of @xmath424 follows from the fact that @xmath658 and @xmath659 can be chosen as continuous maps , because of the existence of ( continuous ) linear right inverses of surjective linear maps to finite dimensional vector spaces .
we consider the special case where @xmath136 is a compact manifold and @xmath660 the frchet algebra of all smooth real - valued functions on @xmath136 . according to @xcite or @xcite , the universal topological differential module of @xmath0
is given by @xmath661 , the space of smooth @xmath662-valued @xmath106-forms on @xmath136 , and the de rham - differential @xmath663 is a universal continuous derivation .
it follows in particular that the space @xmath189 is the space of exact @xmath106-forms , which is non - zero .
now let @xmath1 be a finite - dimensional real lie algebra and @xmath664 up to cocycles vanishing on @xmath330 , all continuous cohomology classes in @xmath665 are then represented by sums @xmath361 , where @xmath666 is an alternating continuous linear map for which there is a continuous linear map @xmath667 and @xmath668 we interprete the continuous linear map @xmath669 as a @xmath449-valued _ current on @xmath136_. it is a closed current if and only if it vanishes on exact forms .
typical examples of such currents arise from pairs @xmath670 , where @xmath671 \to m$ ] is a piecewise smooth path and @xmath672 via @xmath673 but these examples satisfy @xmath674 .
the following section is very much based on information and hints we got from m. bordemann ( @xcite ) .
let @xmath1 be a lie algebra and @xmath305 a @xmath1-module .
we denote the action as @xmath675 by @xmath676 . on the space @xmath677 of @xmath305-valued
lie algebra cochains we have a natural action of @xmath1 denoted by @xmath678 , x_{i+1 } , \ldots , x_p).\ ] ] for @xmath679 , we consider the injection @xmath680 from the action of @xmath1 on the spaces @xmath681 , we obtain lie algebra differentials @xmath682 and we also have @xmath683 satisfying on @xmath684 the identity @xmath685 ( cf .
@xcite , lemma 1 ) .
specializing to the trivial module @xmath686 , we obtain in particular the maps @xmath687 which , in view of equation ( [ eq : homform ] ) , commute with the respective lie algebra differentials because @xmath688 vanishes on @xmath689 .
hence they induce linear maps @xmath690 \mapsto [ \tilde \alpha_p(\omega)].\ ] ] for the @xmath1-module @xmath691 , the subspace @xmath692 of @xmath693 consists of maps whose associated bilinear map is alternating .
we thus have a well - defined map @xmath694 which is a morphism of @xmath1-modules .
we now obtain maps @xmath695 satisfying @xmath696 hence @xmath697 induces a linear map @xmath698 from the construction , we immediately get @xmath699 , which leads to @xmath700 . to see that for each cocycle @xmath703 the symmetric bilinear form @xmath704 is invariant
, we note that @xmath705,z ) = \omega([x , y])(z ) + \omega(z)([x , y]),\ ] ] and if @xmath706 is a cocycle , this can be written as @xmath705,z ) = ( x.\omega(y))(z ) - ( y.\omega(x))(z ) + \omega(z)([x , y ] ) = \omega(y)([z , x ] ) + \omega(x)([y , z ] ) + \omega(z)([x , y]),\ ] ] showing that this trilinear form is alternating , and hence that @xmath704 is invariant .
* exactness in @xmath707 : * we only have to show that @xmath708 is injective .
if @xmath709 satisfies @xmath710 for some @xmath711 , then @xmath712),\ ] ] which implies that @xmath706 is a @xmath9-coboundary .
* exactness in @xmath713 : * clearly @xmath714 .
if , conversely , @xmath715 ) = 0 $ ] , then @xmath716 is a linear map whose associated bilinear form @xmath717 is alternating .
in this situation , we have @xmath718)(z ) - \omega([y , z])(x ) - \omega([z , x])(y ) = -\omega([x , y])(z ) + \omega(x)([y , z ] ) + \omega(y)([z , x ] ) \cr & = & \big(- \omega([x , y ] ) - y.\omega(x ) + x.\omega(y)\big)(z ) = ( d_\k \omega)(x , y)(z).\end{aligned}\ ] ] we conclude that @xmath719 is a cocycle if and only if @xmath706 is one , and from that we derive that @xmath720 .
* exactness in @xmath449 : * next we show that @xmath721 .
so let @xmath703 and write @xmath722 , where @xmath723 is symmetric and @xmath724 is alternating . then @xmath725)(z )
+ \omega(z)([x , y ] ) = \omega(y)([z , x ] ) + \omega(x)([y , z ] ) + \omega(z)([x , y ] ) , \end{aligned}\ ] ] and the closedness of @xmath706 also shows that @xmath726)(z ) = 2 \sum_\cyc \omega(x)([y , z]),$ ] which leads to @xmath727)(z ) - \omega(z)([x , y ] ) = 2 \sum_\cyc \omega_-([x , y],z ) = - 2 d_\k \omega_-(x , y , z).\ ] ] hence @xmath728 is always exact , so that @xmath729 vanishes on the level of cohomology spaces . to see that @xmath730 , suppose that @xmath72 is an exact invariant bilinear form and @xmath731 satisfies @xmath732
. then @xmath733 defines a linear map @xmath716 with @xmath734))(z ) \cr & = & \omega(y)([z , x ] ) + \omega(x)([y , z ] ) - \omega([x , y])(z ) \cr & = & \kappa([x , y],z ) + \eta(y,[z , x ] ) + \eta(x,[y , z ] ) - \eta([x , y],z ) \cr & = & \kappa([x , y],z ) + d_\k\eta(x , y , z ) = ( \gamma(\kappa ) + d_\k\eta)(x , y , z ) = 0.\end{aligned}\ ] ] from the preceding calculation , we also see by putting @xmath735 , that the linear map @xmath736 defined by an invariant symmetric bilinear form @xmath72 is a @xmath106-cocycle if and only if @xmath77 vanishes . *
exactness in @xmath532 : * the transfer formula for differentials implies that an alternating trilinear form @xmath706 on @xmath1 is a @xmath73-cocycle if and only if the corresponding alternating bilinear form @xmath737 is a @xmath9-cocycle
. therefore the image of @xmath738 consists of those cohomology classes having a representing cocycle whose associated trilinear form is alternating .
for @xmath76 , the corresponding @xmath73-cocycle @xmath77 and the corresponding linear map @xmath736 , we have @xmath739,\cdot ) = - ( d\tilde\kappa)(x)(y)\ ] ] because @xmath740)(z ) \cr & = & -\kappa(y , [ x , z ] ) + \kappa(x,[y , z ] ) - \kappa([x , y],z ) = \kappa([x , y],z ) .
\end{aligned}\ ] ] we conclude that @xmath741 is exact , so that @xmath742 induces the trivial map @xmath743 .
let @xmath744 and write @xmath745 .
we then have @xmath746))(z ) = f(y)([z , x ] ) + f(x)([y , z ] ) - f([x , y])(z ) \cr & = & f(y)([z , x ] ) - ( y.\tilde f)(x , z ) .
\end{aligned}\ ] ] this map is alternating in @xmath747 , and it is alternating in @xmath748 if and only if @xmath749 is alternating . writing @xmath750 for the decomposition of @xmath751 into symmetric and alternating components ,
this is equivalent to @xmath752 .
we conclude that @xmath753 is alternating if and only if @xmath754 is invariant .
to verify the exactness in @xmath532 , we now assume that @xmath755 satisfies @xmath756 , i.e. , @xmath757 ) = 0 $ ] .
then there exists an @xmath744 with @xmath758 , and the preceding paragraph implies that @xmath754 is an invariant symmetric bilinear form on @xmath1 satisfying @xmath759 where @xmath760 corresponds to the decomposition @xmath761 . we conclude that @xmath762 = -[\gamma(\tilde f_+)]$ ] , which implies exactness in @xmath532 . *
exactness in @xmath763 : * we claim that @xmath764 .
to verify this claim , pick @xmath765 for which @xmath766 is exact , i.e. , there exists a symmetric bilinear form @xmath767 with @xmath768 , i.e. , for @xmath769 we have @xmath770,z ) - \kappa(y,[x , z]).\ ] ] let @xmath771 and write @xmath772 for the corresponding bilinear map on @xmath1 with @xmath773 . then @xmath774))(z )
= -\eta(y,[x , z ] ) + \eta(x,[y , z ] ) - \eta([x , y],z).\ ] ] therefore @xmath775 ) - \tilde\eta([x , y])(z ) + \tilde\eta(z)([y , x ] ) - \tilde\eta([x , z],y ) \cr & = & 2 ( \tilde\eta_+([y , x])(z ) + \tilde\eta_+([z , x])(y ) ) = 2 ( x.\eta_+)(y , z),\end{aligned}\ ] ] and this leads to @xmath776 since @xmath777 , we find some @xmath778 with @xmath779 , and then @xmath780 so that for @xmath781 the corresponding trilinear map is alternating .
this means that @xmath782 = [ \omega ' ] \in \im \alpha_3 $ ] .
p. maier , _ central extensions of topological current algebras _ , in `` geometry and analysis on finite- and infinite - dimensional lie groups '' , a. strasburger et al eds .
, banach center publications * 55 * , warszawa 2002 ; 6176 k .- h .
neeb , _ current groups for non - compact manifolds and their central extensions _ , in `` infinite dimensional groups and manifolds '' . edited by tilmann wurzbacher .
irma lectures in mathematics and theoretical physics * 5 * , de gruyter verlag , berlin , 2004 ; 109183 | let @xmath0 be a unital commutative associative algebra over a field of characteristic zero , @xmath1 be a lie algebra , and @xmath2 a vector space , considered as a trivial module of the lie algebra @xmath3 . in this paper
, we give a description of the cohomology space @xmath4 in terms of well accessible data associated to @xmath0 and @xmath1 .
we also discuss the topological situation , where @xmath0 and @xmath1 are locally convex algebras .
+ keywords : current algebra , lie algebra cohomology , lie algebra homology , invariant bilinear form , central extension + ams - class : 17b56 , 17b65 |
the nucleus of female mammals embryo cells contain two x chromosomes ; one of these has to be inactivated in order to equalize the dosage of x genes product with respect to males , and this inactivation is crucial to survival @xcite .
moreover , the x chromosome is conjectured to play a key role in the arising of certain types of cancer @xcite .
the mechanism by which one of the two x chromosomes is inactivated is poorly understood , despite extensive work on this problem @xcite ; it is known that co - localization of chromosomes at the time x - chromosome inactivation ( xci ) is established plays a key role @xcite .
very recently , nicodemi and prisco @xcite proposed a model to explain xci on the basis of a spontaneous symmetry - breaking mechanism , related to the classical ising model of statistical mechanics , in a dynamical stochastic model of the single - agent type .
their idea is that blocking factor ( bf ) molecules diffuse in the cell and can bind randomly to one or the other of the x - chromosomes ; however , by a collective effect , the affinity to a chromosome increases when this already binds to other bf molecules .
the net effect is that once one of the two chromosomes , by the effect of a fluctuation , binds to a larger number of bf molecules , it will bind to more and more of these at a rate higher than the other chromosome ( note that the bf molecules could have a greater probability to escape when more of them bind to the chromosome ; this effect should be dominated by the growth in affinity in order for the collective effect devised by nicodemi and prisco to set in ) . in the whole , as a result of diffusive behavior , fluctuation and the affinity - enhancing collective effect , a net current will be established leading to bf concentration on one of the two x - chromosomes . the model can also account for the relevance of co - localization at the establishment of xci ; in facts , if the two x - chromosomes are too far away , the diffusive behavior will prevail , and the bf molecules escaping from the less populated x - chromosome will have a very low probability to reach the other one .
the purpose of this short note is to provide a deterministic ( simplified ) version of the nicodemi - prisco symmetry - breaking xci model , considering average exchanges of blocking factors between each of the two x - chromosomes and the cell fluid environment .
thus our model will amount to a mean - field version of the nicodemi - prisco model ; we will formulate it first in terms of transition probabilities between states for a single bf molecule , and then pass to a formulation in terms of a first order system of odes , i.e. a dynamical system @xcite , following a procedure which is standard in physics or in quantitative biology @xcite . here
we will consider the fluid environment as a homogeneous reservoir , i.e. overlook the effects connected with the spatial localization of the x - chromosomes ; thus the model will be able to predict the symmetry breaking leading to inactivation of one of the x - chromosomes , but ( in its simplest form , presented here ) not its dependence on co - localization of the chromosomes .
an extension of the model aiming at including co - localization effects will be developed in forthcoming work .
similarly , here we will only deal with average quantities and flows ; fluctuations can be included within this kind of modeling by promoting the odes we obtain to _ stochastic differential equations _
@xcite , as briefly discussed below .
we will consider a fixed total amount ( @xmath0 units ) of bf molecules ; these can be in three mutually exclusive states : they can bind to one ( state 1 ) or the other ( state 2 ) of the two x - chromosomes , or be diffusing in the cell environment ( state 3 ) .
we denote the number of units binding to the first x - chromosome at time @xmath1 as @xmath2 , that of units binding to the second x - chromosome at time @xmath1 as @xmath3 , and that of units diffusing in the cell fluid at time @xmath1 as @xmath4 . we will refer to @xmath5 and @xmath6 as the _ occupation numbers _ for the two x - chromosomes ( or more precisely , for the xic of the chromosomes ) .
obviously , [ totnumber ] x(t ) + (t ) + z(t ) = n .
let us consider the probabilities @xmath7 for transitions between these states in particular , transition from state @xmath8 to state @xmath9 for each bf molecule ( referred to as a `` particle '' in the following ) in a time interval of length @xmath10 . in the following , we will use a simplified notation , and write [ probshort ] p_ij : = p(i j , t ) .
we assume that for @xmath11 sufficiently small , the probability a particle initially in state 3 will end up being in state 1 or 2 ( i.e. bound to one of the two x - chromosomes ) is [ prob1 ] p_31 ( x ) t + o ( t ) ; p_32 ( ) t + o ( t ) .
that is , it will be proportional to the length of the time interval ( but independent of time @xmath1 itself ) via a function @xmath12 or @xmath13 representing the affinity with the concerned x - chromosome and depending on the number of particles already binding to it .
similarly , the probability that a particle initially binding to one of the x - chromosomes will escape from it in a time interval @xmath11 will be given by [ prob2 ] p_13 ( x , z ) t + o ( t ) ; p_23 ( , z ) t + o ( t ) .
that is , the escape probability will be proportional to the length of the time interval via a function of the number of particles binding to the concerned x - chromosome and of the number of particles fluctuating in the fluid environment ( it is natural to assume these should enter only through their density @xmath14 ) .
as for direct transitions from one to the other of the two x - chromosomes , these will be impossible on account of their non - zero space separation ( the particle will have to diffuse through the fluid to do that ) , so we will set [ prob3 ] p_12 = o ( t ) ; p_21 = o ( t ) .
needless to say , the probabilities of permanence in a given state are obtained simply requiring the conservation of particles , i.e. that the sum of probabilities for all transitions from a given state to all possible state is one ( markov property ) ; thus [ prob4 ] rl p_11 = & 1 - p_12 - p_13 = 1 - ( x , z ) t + o ( t ) , + p_22 = & 1 - p_21 - p_23 = 1 - ( , z ) t + o ( t ) , + p_33 = & 1 - p_31 - p_32 = 1 - [ ( x ) + ( ) ] t + o ( t ) .
let now @xmath15 be the average occupation numbers for states @xmath16 at time @xmath1 ; we look at the average occupation numbers at time @xmath17 .
note @xmath18 can be expressed using , hence we need only two equations .
these will be given by [ eveq1a ] rl x ( t+ t ) = & x(t ) - p_13 x(t ) + p_31 z(t ) , + y ( t+ t ) = & y(t ) - p_23 y(t ) + p_32 z(t ) . using the expressions given above for @xmath19 and omitting @xmath20 terms the yield [ eveq1 ] rl x ( t+ t ) = & x(t ) + [ ( x(t ) ) z(t ) - ( x(t),z(t ) ) x(t ) ] t + y ( t+ t ) = & y(t ) + [ ( y(t ) ) z(t ) - ( y(t),z(t ) )
y(t ) ] t . dividing by @xmath11 and letting @xmath21 , eq .
yields a system of odes describing the evolution of the average occupation numbers for the three states ( in which we make explicit the expression of @xmath18 implied by the conservation law ): [ eveq3 ] rl d x / dt = & ( x ) ( n - x - y ) - ( x , n - x - y ) x , + d y / dt = & ( y ) ( n - x - y ) - ( y , n - x - y ) y .
passing to consider the densities @xmath22 and @xmath23 , and introducing the functions @xmath24 and @xmath25 defined through a ( w ) : = ( n w ) , b(w_1,w_2 ) : = ( n w_1,n w_2 ) , we get in the end the equations ( symmetric under the exchange of @xmath26 and @xmath27 ) [ eveq4 ] l d x / dt = a(x ) ( 1-x - y ) - b(x,1-x - y ) x , + d y / dt = a(y ) ( 1-x - y ) - b(y,1-x - y ) y .
so far we have worked in completely general terms within the three - states description of the system . in order to have a specific model , we should choose concrete forms for the functions @xmath28 and @xmath29 and hence of the @xmath30 and @xmath31 describing the probability of capture by and escape from chromosomes in function of the associated occupation numbers @xmath32 or densities @xmath33 ; this will yield a specific expression for the evolution equations and .
in their work , nicodemi and prisco @xcite argued that the essential feature of their model for xci is the collective phenomenon enhancing affinity and hence the probability of capture by x - chromosomes with a substantial number of binding particles .
thus the function @xmath34 should go to a finite limit @xmath35 for @xmath36 ( the bf have a non - zero affinity with the chromosomes even when no bf molecules are binding to chromosomes , and more generally when the collective behavior has not set in ) , and grow with @xmath37 , i.e. @xmath38 . as for @xmath39
, we should similarly have a function with a finite limit @xmath40 for @xmath36 ; as for its dependence on @xmath37 and @xmath41 , we assume it is only through the ratio @xmath42 of bf molecules binding to the chromosome and bf molecules in the fluid environment . in the following ,
we choose to deal with a
minimal model , i.e. consider a linear dependence on @xmath43 for both these quantities ; we set [ ab ] ( w ) = a_0 + a_1 w , ( w ) = b_0 + b_1 ( w / z ) ; this implies of course a ( w ) = a_0 + a_1 n w , b ( w ) = b_0 + b_1 ( w / z ) .
the parameters @xmath44 are all positive ( we could always set one of these parameters equal to one by rescaling the time unit . ) we also stress that in view of the discussion by nicodemi and prisco @xcite , we should expect the cooperative effects enhancing affinity should be predominant over those depressing the escape rate ; this means we should expect @xmath45 rl d x / d t = & ( a_0 + a_1 n x ) ( 1 - x - y ) - [ b_0 + ( b_1/(1 - n(x+y ) ) ) n x ] x , + d y / d t = & ( a_0 + a_1 n y ) ( 1 - x - y ) - [ b_0 + ( b_1/(1 - n(x+y ) ) ) n y ] y .
we are specially interested in the large @xmath46 regime . in order to study this , it is convenient to rescale time ( note that eqs . become singular for @xmath47 ) and pass to consider as independent variable : = n t .
we also set : = 1/n ( hence @xmath48 ) ; the equations read then [ eqseps ] rl d x / d = & a_1 ( 1 - x - y ) x + [ a_0 ( 1 - x - y ) - b_0 x ] + b_1 x^2 ( x+y - ) ^-1 , + d y / d = & a_1 ( 1 - x - y ) y + [ a_0 ( 1 - x - y ) - b_0 y ] + b_1 y^2 ( x+y - ) ^-1 ; at first order in @xmath49 , we have [ eqslim ] rl d x / d = & a_1 ( 1 - x - y ) x + [ a_0 ( 1 - x - y ) - ( b_0 - b_1 x /(x+y ) ) x ] , + d y / d = & a_1 ( 1 - x - y ) y + [ a_0 ( 1 - x - y ) - ( b_0 - b_1 y /(x+y ) ) y ] .
we want now to discuss some point concerning the qualitative behavior of the dynamical system defined by ; we should consider it in the region ( invariant under the dynamics ) of @xmath50 delimited by the coordinate axes and by the line @xmath51 . [ cols= " < , < " , ] obviously dominance of fluctuation in the dynamics will take place once the system gets near or very near if @xmath52 is small to the stable equilibrium point @xmath53 or @xmath54 .
there will be a metastable equilibrium distribution around @xmath55 and over extremely long times
large fluctuations could drive the system near to the other stable equilibrium ; the equilibrium distribution and the time needed for large fluctuations to appear can be estimated via standard stochastic processes techniques @xcite . in figure [ fig4 ]
we show a numerical simulation of ( the stochastic process defined by ) the stochastic differential equations .
we have provided a mean - field formulation of the model recently proposed by nicodemi and prisco @xcite for the x - chromosome inactivation in mammals @xcite . like their model
, the version proposed here explains @xmath56-chromosome inactivation as the result of a _ spontaneous symmetry breaking _ in the dynamics of blocking factors molecules binding to the x - chromosome inactivation centers ( xic ) .
while their approach was based on a single - agent model , in our version the considered equations deal directly with average quantities . after discussing a general formulation of the model
, we considered a minimal version , in which the affinity grows linearly with the number of bf molecules already binding to the xic and the rate of escape of bf molecules increases linearly on the ratio between density of bf molecules binding to the x - chromosomes and of bf molecules floating in the fluid environment with the same number .
our model contains four positive parameters , related to affinity and escape probability for the bf molecules .
we showed that when @xmath57 , non - symmetric equilibria exist ; under the same condition they are stable ( while symmetric equilibria are unstable ) .
thus in such case we have a _ spontaneous symmetry breaking _ , i.e. non - symmetric equilibria describe the asymptotic state of the system .
we were also able to estimate the rate of approach to equilibrium as a function of the parameters describing the system and the total bf population .
a qualitative discussion showed that when the initial situation corresponds to most of the bf molecules floating in the fluid , one should expect the dynamics to undergo different phases .
we ran some numerical simulations ( one of these is shown in the figures ) confirming the conclusions reached by the qualitative discussion mentioned above .
we also briefly discussed how fluctuations can be taken into account within our scheme .
finally , we would like to stress that our model differs from that of nicodemi and prisco only in the mathematical description it considers , while the physical basis is the same .
thus , we are just providing an alternative description in terms of dynamical systems and odes rather than of stochastic processes for the physics described by nicodemi and prisco in their paper @xcite .
this mathematically simpler description allows to make more detailed predictions , in particular about different timescales in the dynamics ; these were fully confirmed by numerical simulations .
our model , like the one of nicodemi and prisco @xcite , did not take into account co - localization ; work is in progress to take this into account within the present description of x - inactivation .
in this appendix we will briefly discuss the passage from the physical model of section [ sec : model ] to the stochastic differential equations and . in order to take into account full detail of fluctuations in our model
, we should consider the flows @xmath58 of particles from state @xmath8 to state @xmath9 . denoting by @xmath59 the number of particles passing from state @xmath8 to state @xmath9 in a time interval of length @xmath60 , we have of course _ ij = _ ij - _ ji .
the transition probabilities @xmath19 for a single particle are given in section [ sec : mod : tp ] ; the @xmath61 will thus be poisson distributed with parameter @xmath62 , where @xmath63 is the population of state @xmath8 at the beginning of the time interval .
needless to say , if we look just at expectation values we obtain again ; however , our present task is to go beyond this level of description , and take into account first and higher order momenta of the poisson distribution that is , in physical terms , _ fluctuations _ around the average dynamics .
let us fix our attention , for the sake of concreteness , on the population @xmath56 ( or density @xmath22 ) of the state 1 and its variation @xmath64 ( or @xmath65 ) in a time interval of length @xmath10 ; we also write @xmath41 and @xmath66 rather than expressing these quantities in terms of @xmath67 and @xmath68 , for ease of notation .
we obviously have @xmath69 .
recall that @xmath70 is poisson distributed with parameter @xmath71 ( i.e. @xmath72 ) , and @xmath73 is also poisson distributed but with parameter @xmath74 .
thus @xmath75 is poisson with mean @xmath76 and variance @xmath77 .
thus for @xmath56 we have the stochastic differential equation @xmath78 ; in the case of our minimal model this reads d x = [ ( a_0 + a_1 x ) z - ( b_0 + b_1 x / z)x ] d t + d ( t ) . passing to density variables this reads @xmath79 d t + n^{-1}[(a_0 + a_1 n x ) - ( b_0 + b_1 x / z ) n x ] ^{1/2 } d \om(t)$ ] ; we should also pass to the rescaled time variable @xmath80 , so that @xmath81 and @xmath82 .
in this way we get , using again @xmath83 , d x = [ a_1 x z + ( a_0 z - ( b_0 + b_1 x / z ) x ] d + d ( ) .
we thus have a noise of intensity @xmath84 ; note that at order zero in @xmath49 , @xmath85 ; as the sum of @xmath26 and @xmath66 can not be higher than one , this can be estimated by @xmath86 . | nicodemi and prisco recently proposed a model for x - chromosome inactivation in mammals , explaining this phenomenon in terms of a spontaneous symmetry - breaking mechanism [ _ phys .
rev .
lett .
_ 99 ( 2007 ) , 108104 ] . here
we provide a mean - field version of their model .
[ firstpage ] |
simulation of organic semiconductor devices , e.g. organic light emitting diodes ( oled ) , organic solar cells , etc . , has gained great interest in the past decade .
accurate models often lack in precise parameters and their identification is time - consuming or expensive .
one major problem is the uncertainty in the measurement data which leads to uncertain parameters . carrying out more experiments
would minimize this uncertainty in an expensive way .
an alternative is to use oed in order to minimize the parameter uncertainty by planning new ( optimal ) experiments .
new measurement data are received for which the parameter estimation yields parameters with minimal confidence intervals .
we apply the concept of oed to the egdm , a special model for the mobility of electron transport in organic polymeric material .
the chapters are arranged as follows : we give a brief overview of the model equations in sec . [ sec:2 ] and point out what the relevant quantities are . in sec .
[ sec:3 ] , we explain the methodology of the optimum experimental design problem in more detail . after describing the equation solver methods we used , sec .
[ sec:4 ] , numerical results of the optimization are presented in sec .
[ sec:5 ] .
a basic description of charge transport in semiconducting materials in the steady - state case is given by the coupled van roosbroeck system@xcite consisting of the continuity equation , also called drift - diffusion equation , and the poisson equation . given a domain @xmath0 , the state variables , i.e. the space dependent functions , are the electric charge density @xmath1 in @xmath2 $ ] and the electric potential @xmath3 in @xmath4 $ ] which are scalar real valued functions defined on @xmath5 .
we assume that @xmath1 and @xmath3 are twice differentiable on @xmath5 .
pasveer et al .
@xcite proposed the egdm to conjugated semi - conducting polymers , where the diffusion and the mobility depend on the state variables and hence on the space variable @xmath6 .
furthermore they introduced another state @xmath7 , called quasi - fermi - energy , and a corresponding equation which couples @xmath7 and @xmath1 at every space point .
we omit the @xmath6-dependence in the following equations , only @xmath1 , @xmath3 and @xmath7 are space dependent .
the model equations are : @xmath8 where @xmath9 , \\ g_2(\phi ) & = \exp\left\ { 0.44(\hat{{\sigma}}^{\frac32}-2.2)\right\ } \\ & \qquad \left[\sqrt{1 + 0.8\left ( \min\left\{\frac{\partial_x \phi}{{n_t}^{\frac13}{\sigma}},2\right\}\right)^2 } -1\right ] , \\ g_3(n , e_f ) & = \frac{n}{k_b { t } \frac{dn}{de_f}}. \label{eq : g3 } \end{aligned}\]]in these equations @xmath10 is the electric current density in @xmath11 $ ] , @xmath12 is the boltzmann constant in @xmath13 $ ] , @xmath14 the permittivity in @xmath15 $ ] , @xmath16 the elementary charge in @xmath17 $ ] , @xmath18 and @xmath19 . in the anorganic case , the @xmath20-factors , @xmath21 , would be constant .
their organic model is yield by comparison with the solution of the master equation .
@xcite on the boundary @xmath22 the following conditions are imposed : @xmath23^{-1}{{\,de } } , \\
n(l ) & = \frac{{n_t}}{\sqrt{2\pi{\sigma}^2 } } \int \limits_{-\infty}^\infty e^{-\frac{e^2}{2{\sigma}^2 } } \frac{1}{1+\exp\left(\frac{e+{\varphi_2 } } { k_b{t}}\right)}{{\,de } } , \\
\phi(0 ) & = 0 , \\ \phi({l } ) & = ev-\left ( \varphi_2 - \varphi_1 \right ) , \end{aligned}\]]where @xmath24 is the voltage in @xmath25 $ ] and @xmath26 are given energy barriers in @xmath4 $ ] . on the entrance , @xmath27 , the energy barrier @xmath28 is lowered according to the theory of emtage and odwyer@xcite and scott and malliaras . @xcite for our computations we take the dimensionless form of the equations proposed by bonham and jarvis . @xcite for the later use we define @xmath29 where @xmath30 are parameters of unknown numerical value given by nature .
they have to be identified by comparing a model response to experimental data .
@xmath31 is the zero temperature mobility in @xmath32 $ ] , @xmath33 is the width of the gaussian distribution of the density of states in @xmath4 $ ] and @xmath34 is the site density in @xmath2 $ ] .
we assemble quantities which are adjustable by an experimenter , the device length @xmath35 in @xmath36 $ ] and the temperature @xmath37 in @xmath38 $ ] , in the vector @xmath39.@xcite
there are two established ways for solving the system -:@xcite 1 .
apply newton s method to the fully coupled system of equations 2 .
use gummel s method , @xcite i.e. a fixed point iteration which decouples the three egdm equations with newton s method , one can achieve quadratic convergence .
however , finding good starting values is not simple . in the work of knapp et al . , @xcite a strategy motivated by physical considerations is described .
another aspect is that the jacobian , i.e. derivatives of the functions w.r.t .
the states , is required . to compute the jacobian , the main two options are to use difference formulas or compute the exact derivatives .
the latter can be done by hand or by using automatic differentiation ( ad ) .
either way , the additional effort for computing the derivative in @xmath1 directions is at least @xmath40 times the effort of each function evaluation in every newton step .
another possibility is to solve - with gummel s method .
table [ tab : gummel ] shows the modified algorithm consisting of the classical system of equations and and the quasi - fermi energy @xmath7 defining equation .
@xcite we solve the equations sequentially and insert the interim results into the next equations .
.[tab : gummel]algorithm for the fixed point iteration of gummel expanded by the egdm - quasi - fermi - equation and a derivative - free linearization .
[ cols= " < " , ] with the special linearization , used in the third equation of table [ tab : gummel ] , we do not need any derivatives and the effort of function evaluations is limited .
we discretize the infinite dimensional problem - to a finite one . the domain @xmath41 is divided in @xmath42 subintervals @xmath43,\ x_i\in \{x_0=0,\dotsc , x_n = l\}$ ] with a constant mesh size @xmath44 .
finite differences are applied to the spatial derivatives , i.e. with respect to @xmath6 .
the so - called scharfetter - gummel scheme@xcite forces the function @xmath10 , defined in eq .
, to be constant on each interval @xmath45 , denoted by @xmath46 , and provides an upwind stabilization , so that computation on coarse meshes is possible . on the interval @xmath45 , the scheme looks like @xmath47 the terms @xmath48 , @xmath49 stand for average values of the non - constant functions @xmath50 ,
it is important that the averages are taken in an upwind conform way to prevent numerical oscillations .
in this part , we follow the approaches of lohmann @xcite and krkel et al .
@xcite with different choices of the controls @xmath39 and the voltage @xmath24 , defined in sec . [ sec:2 ] , we set up multiple experiments in which the current density @xmath10 is measured .
let @xmath51 be the number of measurements we yield . in a parameter estimation ,
the parameters are identified by fitting a model response , here @xmath10 , to experimental data , i.e. measurements .
we assume the measurement error to be normally distributed with mean zero and covariance matrix @xmath52 . with the same experimental settings , i.e. equal controls @xmath39 ,
a fit from a different realization of the measurement errors may result in very different parameter values .
the covariance matrix of the parameters allows to analyze the quality of a parameter estimation .
the assumed model for the standard deviations of the measurement errors is : @xcite @xmath53,\ ] ] where @xmath54 is the function value of @xmath10 corresponding to the @xmath55-th measurement . for further notation
, we assemble the values @xmath56 in the vector @xmath57 .
if the confidence region of the parameters is approximated by assuming a linear propagation of the measurement errors , it can be parameterized by the covariance matrix defined by @xmath58)(p-\mathbb e[p])^t ] \in \mathbb r^{n_p\times n_p},\ ] ] where @xmath59 is the number of parameters
. from now on , we denote by @xmath1 , @xmath3 and @xmath7 the discrete counterparts of the state variables which are @xmath60-dimensional vectors , without boundary values .
they assemble the function values at the mesh points @xmath61 , @xmath62 , cf .
sec.[sec:3 ] .
we abbreviate the discrete solution of the system - dependent on parameters @xmath30 and controls @xmath39 by @xmath63 cf .
table [ tab : gummel ] .
we used the notation @xmath64 for the overall state dimension . in the following
we denote the derivative of a function @xmath65 w.r.t .
@xmath6 in the direction @xmath66 by @xmath67 and write accordingly the second derivatives of @xmath65 w.r.t . @xmath6 and @xmath68 in the directions @xmath66 and @xmath69 as @xmath70 .
we combine several directions @xmath71 into a so - called seed matrix @xmath72 and define @xmath73 by @xmath74 and accordingly @xmath75 by @xmath76 with @xmath77 and @xmath78 .
we define the @xmath79 matrix @xmath80 , \end{aligned}\ ] ] with the derivative of @xmath81 w.r.t . to the parameters @xmath30 with the @xmath59-dimensional identity matrix @xmath82 as seed matrix .
computing @xmath83 is much less expensive than computing the matrix product @xmath84 with the identity matrix @xmath85 .
the covariance matrix in the unconstrained case can be computed by @xmath86 for given probability @xmath87 $ ] the linearized @xmath88-confidence region is described by @xmath89 with the @xmath90-quantile of the @xmath91-distribution @xmath92 . as an approximation of the confidence intervals of the parameters we use @xmath93 \quad \text{with } \\ & \theta_i = \gamma(\alpha)\sqrt{({\operatorname{cov}_p})_{ii } } , \quad i=1,\dotsc , n_p .
\end{aligned}\ ] ] we point out that for computing @xmath94 ( and the confidence intervals ) no measurement data is necessary .
we consider the reduced nonlinear optimization problem @xmath95 which is called `` optimum experimental design problem '' .
efficient optimization algorithms require derivative information . for higher order difference formulas ,
the number of function evaluations increases which can be very costly like in our case .
additionally , low order schemes reduce the number of exact digits , in particular in combination with derivatives of higher order . to avoid such approximation errors , we use automatic differentiation where the derivatives are evaluated up to machine precision . for derivative based optimization
, we need the derivative of the gradient of @xmath96 with the seed matrix @xmath97 , where @xmath98 is the number of the controls . @xmath59 and @xmath98 are relatively small , so we use the forward mode , where we have to compute @xmath99 \\ & \ = -\sigma^{-1 } \big [ \partial^2_{u}\mathcal j\{\partial_p u \{i_p\ } , \partial_q u \{i_q\}\ } + \partial^2_{qu}\mathcal j\{\partial_p u \{i_p\},i_q\ } \\ & \ \qquad\qquad + \partial_u \mathcal j \big\{\partial^2_{qp}u \{i_p , i_q\ } \big\ } + \partial^2_{up}\mathcal j\{i_p,\partial_q u \{i_q\}\ } \\ &
\ \qquad\qquad + \partial^2_{qp}\mathcal j\{i_p , i_q\ } \big ] , \end{aligned}\ ] ] see e.g. griewank .
@xcite we need derivatives of @xmath100 w.r.t . the parameters @xmath30 , controls @xmath39 and mixed second derivatives w.r.t .
@xmath30 and @xmath39 .
one way would be to differentiate the fixed point iteration , table [ tab : gummel ] , and iterate over the derivatives as well .
@xcite however , we are only interested in the derivatives of the solution of gummel s method @xmath101 . in the solution it holds @xmath102 with @xmath103 the discrete counterpart of the system of equations - . to compute the required derivatives @xmath104 , @xmath105 and @xmath106 , we use the sensitivity method , see e.g. hinze et al .
@xcite using adjoint method is not recommended here , because the number of measurements @xmath51 is much higher than the number of parameters @xmath59 . differentiating leads to @xmath107 hence the required derivatives are given in form of @xmath108 we can save the decomposition of the matrix @xmath109 and use it for all three equations to reduce complexity .
experimenters proceed in the following way.@xcite devices of different lengths @xmath35 are produced by evaporating . at different temperatures @xmath37 , they apply voltage series @xmath110 , a vector consisting of various voltages @xmath24 , and measure the corresponding current values , combined in the vector @xmath81 .
one choice of @xmath35 , @xmath37 and voltage series @xmath110 we call one experiment subsequently .
we will show in two examples how the confidence regions can be reduced by oed . in each example
we consider three different lengths @xmath111 and three different temperatures @xmath112 the combination of each element of the vector @xmath113 with each element of the vector @xmath114 lead to nine experiments .
we assemble all control variables , which are the optimization variables for the oed problem , in a vector @xmath115 the result of the oed should be three lengths and three temperatures combined in nine experiments . to enforce that
, we apply additional constraints to @xmath116 for the lengths @xmath117 and for the temperatures @xmath118 we solve problem with the software package _
vplan_@xcite where the optimization problem is solved with an inexact sqp method provided by _
snopt 7.2 - 9 ( jun 2008)_. @xcite in each iteration , the objective , i.e. the covariance matrix , is computed by solving the underlying system of equations .
we implemented the extended gummel method table [ tab : gummel ] in _ vplan _ to solve this system .
for the optimization , the derivatives of the model functions are calculated with the ad tool _
adifor 2.0_. @xcite we implemented a routine in _ vplan _ , which solves the system of equations . in a first example
we take configurations @xmath30 , @xmath110 , @xmath113 and @xmath114 from pasveer et al .
@xcite who used @xmath119 here and in the following example we set @xmath120 for the energy barriers in @xmath4 $ ] and @xmath121 for the permittivity in @xmath15 $ ] .
we choose boundaries to the lengths and temperatures according to the practicability of experiments and the validity of the model : @xmath122 \quad \text{and } \quad t_i\in[200,350 ] \quad i=1,2,3.\ ] ] the optimization results in the vectors @xmath123 where the lengths do not change much , a tendency to higher temperatures is observed .
the algorithm gives back the exact numbers of the optimum .
dependent on the equipment of the laboratory these values can not be realized in practice .
nevertheless they can be taken as a guideline .
all the values in the neighborhood of @xmath124 and @xmath125 are leading to similar results . in fig .
[ fig:1 ] , we visualize the three dimensional ellipsoid and the projections corresponding to the set @xmath126 before and after the optimization . table [ tab:2 ] shows a comparison of the confidence intervals of the parameters .
the average of the squared semi - axes of the optimized ellipsoid , i.e. the objective in , is @xmath127 times the average of the squared semi - axes of the ellipsoid not optimized .
-confidence regions before ( light part ) and after ( dark part ) the optimization . computed with pasveer parameters . ]
p0.7cmrr + & @xmath128 & @xmath129 + + @xmath130 & 79.90% & 19.85% + @xmath131 & 5.21% & 1.27% + @xmath132 & 21.62% & 7.21% + the results of the simulations before and after the optimization procedure are shown in fig .
[ fig:2 ] .
and @xmath114 and the dashed lines correspond to experiments with optimal configuration @xmath124 and @xmath125 . ]
the optimal configuration leads to higher current densities .
a second example is taken from mensfoort and coehoorn , @xcite who used @xmath133 again we take boundaries like in and the resulting vectors are @xmath134 unlike the first example ,
the optimal lengths differ much more from the starting values and not all temperatures were raised , the lowest temperature was even reduced . in fig . [ fig:3 ]
, we illustrate the confidence ellipsoid and the projections and -confidence regions before ( light part ) and after ( dark part ) the optimization . computed with coehoorn parameters . ]
table [ tab:3 ] shows a comparison of the confidence intervals in this case .
the average of the squared semi - axes of the optimized ellipsoid here is @xmath135 times the average of the squared semi - axes of the ellipsoid not optimized , too .
p0.7cmrr + & @xmath128 & @xmath129 + + @xmath136 & 44.76% & 15.98% + @xmath137 & 7.20% & 3.42% + @xmath138 & 35.42% & 16.51% + the results of the simulations before and after the optimization procedure are shown in fig .
[ fig:4 ] .
the optimized configuration leads to a wider range of current densities gaining more information .
and @xmath114 and the dashed lines correspond to experiments with optimal configuration @xmath124 and @xmath125 . ]
we have applied optimum experimental design to organic semiconductors modeled by the egdm in order to reduce the parameter uncertainty caused by measurement errors .
the variances of the parameters , i.e. the average of the squared semi - axes of the confidence ellipsoid , were @xmath127 and @xmath135 times the average of the squared semi - axes of the not optimized confidence ellipsoid respectively .
the proposed experiments , followed by a parameter estimation lead to a model , which is approximately @xmath139 times more exact with respect to the assumed measurement error than the previous one .
the classical methods for solving semiconductor models were extended to fit with the egdm .
the derivatives required for the optimization were computed via automatic differentiation , leaving the error on machine precision level , even for higher derivatives .
we used unipolar layer devices for simplicity , but the presented methods can also be applied to multi - layer devices , trap generation models and exciton rate equations or , more far - reaching , all other models which are based on the van roosbroeck system .
+ 20ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * ( ) @noop * * ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) , @noop * * ( ) @noop * * , ( ) @noop ( ) , , @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ , no . ( , , ) @noop * * , ( ) http://books.google.de/books?id=gypaygeacaaj[__ ] , mathematical modelling : theory and applications ( , ) @noop ( ) , _ _ , http://www.koerkel.de[ph.d .
thesis ] , , ( ) @noop @noop `` , '' ( , , ) | we apply optimum experimental design ( oed ) to organic semiconductors modeled by the extended gaussian disorder model ( egdm ) which was developed by pasveer et al .
@xcite we present an extended gummel method to decouple the corresponding system of equations and use automatic differentiation to get derivatives with the required accuracy for oed .
we show in two examples , whose parameters are taken from pasveer et al .
@xcite and mensfoort and coehoorn @xcite that the linearized confidence regions of the parameters can be reduced significantly by applying oed resulting in new experiments with a different setup . |
large nonlinear - optical susceptibility materials are central for many optical applications such as telecommunications,@xcite three - dimensional nano - photolithography,@xcite and new materials development@xcite for novel cancer therapies.@xcite in the present work , we focus on the molecular hyperpolarizability , which is the microscopic basis of electro - optic switches and frequency doublers .
the fundamental limits of the hyperpolarizability had been calculated using what appeared to be three - level truncation of the sum rules.@xcite champagne and kirtman were critical of the accuracy of truncated sum rules.@xcite in response , the three - level ansatz was introduced , which states that when a quantum system has a hyperpolarizability near the fundamental limit , the system can be described by a three - level model.@xcite using a new dipole - free form of the sum - over - states expression for the hyperpolarizability,@xcite kuzyk showed rigourously that the three - level ansatz does not demand the sum rules to be truncated.@xcite furthermore , the three - level ansatz with the new dipole free expression was used to calculate the more general fundamental limits of the dispersion of the real and imaginary parts of the hyperpolarizability.@xcite this result can be used to study any phenomena that is based on a second - order response .
more importantly , since the fundamental limit can be determined at any wavelength , it can be used as an absolute standard for comparing hyperpolarizabilities . in particular , the ratio of the measured hyperpolarizability to the fundamental limit , called the intrinsic hyperpolarizability , can be used to compare the efficacy of any two molecules , independent of their size . a factor of thirty gap between the best molecules and the fundamental limit has persisted even after three decades of molecular engineering,@xcite and there is no known explanation for this gap.@xcite it is not of a fundamental nature : a simple clipped harmonic potential energy function , which can be solved analytically , falls well into the gap . as such , a numerical optimization technique , which varies the potential energy function in a way that maximizes the intrinsic hyperpolarizability , was introduced to yield a better understanding of the nature of what makes a large hyperpolarizability.@xcite in that work , zhou and coworkers found that an oscillating wavefunction serves to localize the wavefunctions in a way that forces the system to have only two dominant excited states that interact with the ground state.@xcite this result showed how the three - level ansatz is a natural way for optimizing the intrinsic hyperpolarizability .
based on this observation , zhou and coworkers proposed that molecules with modulation of conjugation , i.e. with wiggly wavefunctions , may be a new paradigm for making molecules with exceptionally large hyperpolarizabilities .
prez moreno and coworkers reported on a series of molecules with varying degrees of modulation in the conjugated bridge between the donor and acceptor ends of these quasi - one - dimensional molecules.@xcite molecules with uneven aromaticity in the bridge , as suggested by zhou s work , were measured to have larger intrinsic hyperpolarizability than those with a smoother bridge .
the best molecule was reported to have a world s record intrinsic hyperpolarizability - 50% better than the previous best .
zhou and coworkers continued to study various ways of optimizing the potential energy function and found that modulated conjugation was but one way of getting near the limit.@xcite depending on the starting potential given to the optimization program , the optimized potential can be wiggly or smooth ; the wavefunctions for the optimized potential can be strongly overlapping or well separated ; and only a few states or many can contribute . in all cases , however , the numerical value of the intrinsic hyperpolarizability is just below 0.71 - about 30% below the fundamental limit .
thus , there appears to be a large number of distinct potential energy functions that yield the same maximum hyperpolarizability .
these results clearly show that while the three - level ansatz appears to accurately the upper bound of the hyperpolarizability , systems that are not well approximated by three - levels can also have an equally large nonlinear response .
but , it is puzzling that all calculated locally maximal values fall exactly 30% short of the limit . in the numerical optimization studies ,
the potential energy function is varied and the transition moments and energies calculated from the resulting wavefunctions .
could the sum rules allow for transition moments and energies that do not result from such simple potential energy functions ? to answer these questions , we take the approach of varying the transition moments and energies directly while enforcing compliance with the sum rules , circumventing the need for calculating wavefunctions from a potential .
this process is numerically more efficient and faster than solving the eigenvalue problem for a hamiltonian , and allows us to investigate a much larger parameter space .
the downside is that we can not easily associate a potential energy function with those matrix elements and energies that yield the largest hyperpolarizability .
we express the sum rules in dimensionless form , @xmath0 where @xmath1 and @xmath2 .
@xmath3 is the energy of state @xmath4 and @xmath5 .
@xmath6 is the electric transition dipole moment between state @xmath4 and @xmath7 ; and @xmath8 is the fundamental limit of the transition moment between the ground and first excited state , where @xmath7 is the mass of the electron , @xmath9 is the magnitude of the electron charge and @xmath10 the number of electrons
. equation [ sumrulenod ] represents an infinite number of equations , one for each distinct value of @xmath7 and @xmath11 .
we can therefore refer to each distinct equation by the pair of indices @xmath12 . in the limit of an @xmath13-level model ( i.e. only @xmath14 excited states contribute to the sos expression for the hyperpolarizability )
, we can treat @xmath15 and @xmath16 for @xmath17 as parameters .
note that we label the ground state with the subscript zero .
we pick the values of these parameters randomly , but constrained by the sum rules , as follows .
we start by defining the energy levels . for an ( @xmath18)-level system ,
we pick @xmath14 random numbers in the range @xmath19 and order these in a vector in ascending order so that @xmath20 .
it is important that @xmath21 since all of the dimensionless energies are given by @xmath22 .
note that @xmath23 and @xmath24 .
once the energies are determined , the sum rules given by equation [ sumrulenod ] are used to fix the transition moments , as follows . beginning with the sum rule equation ( 0,0 ) , we start by picking a random number , @xmath25 , in the range @xmath26 . since each term in the sum in equation ( 0,0 ) is non - negative , @xmath27 .
but , since @xmath28 , this implies that @xmath29 .
we assume that all of the matrix elements of the dipole operator are real , so we set @xmath30 .
equation [ sumrulenod ] with @xmath14 excited states and @xmath31 can be written as : @xmath32 where the numerical value of @xmath33 is computed from the already - determined values of @xmath34 and @xmath35 .
note that @xmath23 , so we drop it . since equation [ sumrulenod0 ] implies that @xmath36 we pick a random number @xmath37 in the range @xmath38 and get @xmath39 using the expression , @xmath40 to get @xmath41 , we evaluate @xmath42 were @xmath43 is another random number .
thus , for the set of random numbers @xmath44 , we continue the process until we have the matrix element @xmath45 .
next , we use the sum rule equation ( 1,1 ) . using the same procedure as above , for a random number @xmath46 , @xmath47 for @xmath48 , @xmath49 and so on , until we get @xmath50 .
we continue the process for sum rules @xmath51 , for all @xmath4 where @xmath52 .
this procedure then yields all of the off - diagonal transition moments .
note that we ignore the sum rule @xmath53 because it gives the nonsensical result that the oscillator strength is negative . with this protocol ,
transition moments of the form @xmath16 for @xmath54 and the energy levels are determined by the role of dice ; but , these parameters are forced to obey the sum rules .
we then use these transition moments and energy levels to calculate the dipole - free intrinsic hyperpolarizability , @xmath55 , which is given by:@xcite @xmath56 , \ ] ] where @xmath57 is the fundamental limit of the hyperpolarizability and @xmath58 is the hyperpolarizability .
note that because the dipole - free expression was derived from the sum rules , and explicitly eliminates dipole moments of the form @xmath59 , we do not need to determine electric dipole moments of the system s eigenstates to calculate the intrinsic hyperpolarizability .
past studies of optimizing the hyperpolarizabilities used the potential energy function as a starting point.@xcite the wavefunctions for a particular potential energy function are determined from the schrodinger equation .
these are then used to determine the dipole matrix and energies of the system , which are in turn the input parameters for calculating the hyperpolarizability using the sum - over - states expression .
the potential energy approach yields several avenues for investigating properties of a system that yield the largest hyperpolarizability .
for example , one can arbitrarily vary the potential energy function to determine directly how the shape of a molecule , characterized by the potential , affects the optical response .
alternatively , one can define a coulomb potential energy function for various arrangements of nuclei to study structure - property relationships.@xcite the latter approach , while more realistic because it more closely describes real molecules , is more limited since it is restricted to a superposition of a finite number of point sources .
the former case , on the other hand , might lead to prescriptions for making artificial systems such as multiple quantum well layered structures or nano - engineered structures .
a more basic question pertains to the limitations imposed by restricting the hamiltonian to be of the form of a kinetic energy term and a potential energy function that depends on position .
it is possible that certain combinations of energy levels and dipole matrices may be derivable only from a more general hamiltonian that , for example , contains terms such as @xmath60 ( i.e. products of functions of the position , @xmath25 , and the momentum @xmath61 ) but that may never - the - less still obey the sum rules .
perhaps these more general hamiltonians fall into classes that each yield a different fundamental limit , but with @xmath62
. our monte carlo method does not discriminate between these different hamiltonians as long as the sum rules are obeyed . alternatively
, matrix elements and energies that are consistent with the sum rules may be inconsistent with some additional basic physical principle .
these are interesting questions that will not be explored here .
rather , we focus on investigating what matrix elements and energy levels lead to an improved nonlinear optical response ; and , how these results compare with those obtained from a simple potential energy function , @xmath63 . .,title="fig : " ] .,title="fig : " ] figure [ fig : distribution]a shows the distribution of the calculated intrinsic hyperpolarizabilities for 10,000 runs using a 5-level model .
first we note that all of the calculated values of the intrinsic hyperpolarizability have a magnitude that is less than unity .
this is consistent with the existence of a fundamental limit .
furthermore , the largest values of @xmath64 approach unity .
in contrast , when the potential energy function is varied,@xcite independent of the starting function , the maximized values converge to about @xmath65 .
so , our results suggest that the hyperpolarizabilities attainable from the simple potential energy functions used in past studies may not be as large as is possible based on only the sum rules .
how the sum rules , which are derived from commutation relationships that assume the potential energy function to be dependent on only the positions of the electrons , yield a different upper bound than direct use of the potential energy function is an open question that will be the focus of future studies .
perhaps such investigations will shed light on why all molecules fall a factor of @xmath66 below the fundamental limit.@xcite if the energy levels are chosen randomly , they will on average be equally spaced . to probe a larger parameter space ,
we define a power law weighting function when randomly choosing the energy level spacing .
thus , on average , the energy level spacing will be of the form @xmath67 .
figure [ fig : distribution]a shows the result for weighting factors with exponents @xmath68 and @xmath69 .
the solid curve shows the function @xmath70 which is fit to the @xmath68 simulation with @xmath4 and @xmath71 as adjustable parameters .
the best fit yields @xmath72 .
1.38% of the values fall outside the range @xmath73 .
figures [ fig : distribution]b and [ fig : distribution]c show the 5-level model results for @xmath74 , and @xmath75 .
the curve given by equation [ eq : distribution ] with @xmath72 is shown for reference . clearly , the simulations with smaller weighting factors , @xmath11 , yield a narrower distribution , showing that the number of simulations giving large intrinsic hyperpolarizability is smaller then the case when @xmath11 is larger .
this is consistent with our recent 3-level model analysis that suggests the second excited state energy should be much higher than the first in order to optimize the hyperpolarizability,@xcite that is , small @xmath76 leads to large hyperpolarizability . since , on average , @xmath77 the intrinsic hyperpolarizability is larger for larger @xmath11 . )
after 100,000 tries for a 10-state model . ] to study the character of the states of a system that leads to the largest nonlinear response , the transition dipole moments and energies that yield the largest value of the intrinsic hyperpolarizability from the set of 100,000 random values generated by our protocol is studied in more detail .
figure [ fig : matrixelements10pm ] shows the matrix elements corresponding to the largest hyperpolarizability(@xmath78 ) in the set for a ten - state model . for this set , the energies , in dimensionless units of the form @xmath79 are : 0 , 1 , 10.08418 , 11.16255 , 17.29753 , 20.40493 , 22.37591 , 27.98258 , 34.93178 , and 45.98839 .
there are several excited states with substantial transition moment , though most of the oscillator strength resides in the lower - lying states .
however , in assessing the contribution of each state to the hyperpolarizability , it is more appropriate to use equation [ dipolefreebeta ] to define the fractional contribution from pairs of individual states @xmath4 and @xmath7 , @xmath80 , \ ] ] , for a 10-state model with @xmath78 ( using matrix elements from figure [ fig : matrixelements10pm ] ) . ]
figure [ fig : betacontribute10pm ] shows a plot of the fractional contribution to the intrinsic hyperpolarizability due to states @xmath4 and @xmath7 .
note that @xmath81 , which is calculated from equation [ dipolefreebetaterms ] , is not symmetric to interchange of @xmath4 and @xmath7 .
however , since @xmath58 is calculated from a sum over terms of the form @xmath82 , the resulting hyperpolarizability is symmetric upon interchange of the indices .
clearly , the two states that contribute most strongly to the intrinsic hyperpolarizability are states 1 and 2 .
state 3 contributes , but is small in comparison .
all other states are negligible .
this clearly shows that when the hyperpolarizability is maximized , the system reduces to a three - state model - consistent with the three - level ansatz .
it is an interesting exercise to assess this large - hyperpolarizability molecule " using an analysis proposed to treat complex molecules in terms of parameters determined from properties of the first two dominant excited states.@xcite in this analysis , the sum rules reduce the sos expression in the three - level ansatz to@xcite @xmath83 where @xmath84 , @xmath85 , @xmath86 and @xmath87{3 } x \sqrt { \frac { 3 } { 2 } \left(1-x^4 \right ) } .\ ] ] equation [ f(e)g(x ) ] should be a good approximation to a molecule whose nonlinear response is close to the limit . with @xmath88 , we get @xmath89 .
the transition moment to the first excited state is given by @xmath90 , which yields @xmath91 .
thus , the intrinsic hyperpolarizability @xmath92 , which is in excellent agreement with @xmath93 when all 10 states are used to calculate the intrinsic hyperpolarizability . note that while many transition moments appear to be substantial ( as we can see from figure [ fig : matrixelements10pm ] ) , the three - level ansatz yields a good approximation . ) after 100,000 tries for a 40-state model ; and ( b ) the fractional contribution , @xmath81.,title="fig : " ] ) after 100,000 tries for a 40-state model ; and ( b ) the fractional contribution , @xmath81.,title="fig : " ] the same pattern holds when more states are included .
figure [ fig : matrixelements40abs]a shows the matrix elements of a 40-state model that yields @xmath94 .
this is the largest value after 100,000 trials . as in the 10-level case ,
we see lots of large matrix elements ; but , the fractional contribution to the intrinsic hyperpolarizability , as shown in figure [ fig : matrixelements40abs]b is dominated by one term corresponding to the first and third excited states .
again , the three - level ansatz appears to hold .
the relevant state are characterized by @xmath95 and @xmath96 , which yields @xmath97 .
this is in reasonable agreement with @xmath98 , which is obtained when all states are included in the sum .
in contrast to the 10-level model , the 40-level model has @xmath99 times more contributions to the hyperpolarizability .
so , when all terms are included there is a larger chance for many smallers terms to add to a significant amount . in all of our calculations , when the number of states are increased , for example , to 80 states and beyond
, the same pattern appears to hold : while many matrix elements are large , only two excited states dominate the intrinsic hyperpolarizability .
figure [ fig : betacontribute40pm - point3 ] shows the contributions to the hyperpolarizability for two systems , with hyperpolarizabilities of 0.033 and 0.31 .
more states contribute as the intrinsic hyperpolarizability gets smaller .
note that for these two cases , similar to the above results , many transition moments are large , but only a few combinations contribute to @xmath58 .
, for a 40-state model when the intrinsic hyperpolarizability is 0.033 ( top ) and 0.31 ( bottom).,title="fig : " ] , for a 40-state model when the intrinsic hyperpolarizability is 0.033 ( top ) and 0.31 ( bottom).,title="fig : " ] to quantify this observation , we define a term in the sos expression for @xmath100 to contribute substantially if its magnitude is at least 10% of the term with the largest contribution .
table [ tab : dominantlevels ] summarizes the dominant states that correspond to the terms that contribute substantially to the hyperpolarizability .
indeed , the number of states that contribute is typically larger when the intrinsic hyperpolarizability is lower .
we stress that this observation is statistically based .
there are certainly cases where the hyperpolarizability can be large even when many states contribute , as we have found in optimization studies of the potential energy function.@xcite however , we conclude that when we randomly pick transition moments and energies , the intrinsic hyperpolarizability is _ usually _ small when many states contribute .
.states whose contributions are at least 10% of the largest contribution for the 40-state model shown in figure [ fig : matrixelements40abs ] . [ cols="<,<,<,<,^",options="header " , ] it is trivial to show from the sum rules that contributions to @xmath58 from states of higher energies can dominate over the lower energy states . however , since the three - level ansatz implies that the higher - energy states are not important when the hyperpolarizability is near the fundamental limit , the higher - lying states must always cancel each other if any of the terms are large . in our protocol
, we force the system to be an n - level model .
we can study the effects of the truncation of the sum rules by considering how the hyperpolarizability changes when we add an additional state . recall that the diagonal sum rules when truncated to n - levels are of the form , @xmath101 where we use the shorthand notation @xmath102 .
equation [ sumrulediagonal ] holds for all @xmath103 , where we find that @xmath104 yields a nonsensical result , so that equation is ignored .
if we add one more state , the diagonal sum rules for @xmath103 are of the form , @xmath105 with the help of equation [ sumrulediagonal ] , equation [ sumrulediagonal+1 ] reduces to @xmath106 assuming that the added state is not degenerate with any of the other @xmath14 states , this yields @xmath107 next we consider the case with @xmath104 .
the sum rule , @xmath108 with the fact that @xmath109 reduces to , @xmath110 so , it must be that @xmath111 and @xmath112 are non - zero and are related to the values of the transition moments and energies that were determined by the monte carlo assignments .
the terms in the dipole - free expression for the hyperpolarizability , given by equation [ dipolefreebetaterms ] , have numerators of the form @xmath113 , where @xmath114 .
thus , the only additional terms that contribute to the hyperpolarizability that include the state @xmath14 are of the form @xmath115 and @xmath116 , where @xmath103 .
both of these terms vanish because @xmath117 according to equation [ zeroxi ] .
if we continue to add more states , we will find that transition moments between these new states and states @xmath14 , @xmath18 , etc . will be nonzero ; but
, the transition moment between the ground state and the newly - added states will vanish because the oscillator strength is depleted by the lowest @xmath14 states .
the net result is that the additional states do not contribute to the hyperpolarizability . since we can continue to add states in this manner , _ ad infinitum _ , to construct a full set of non - truncated sum rules , using truncated sum rules in conjunction with our monte carlo method does not appear to lead to any pathologies when calculating @xmath58
. one might ask if it is physical to have a quantum system in which transition moments from the ground state to all states outside of an n - level subspace vanish .
the harmonic oscillator is an example of a system where transitions from the ground state to all excited states beyond the first vanish .
however , it may be that our monte carlo simulations may yield transition moments and energies that , while consistent with the sum rules , may not be derivable from a potential energy function .
champagne and kirtman had criticized sum - rule truncation , arguing that the resulting limits may be in error.@xcite the later development of the dipole - free sos expression for calculating the hypoerpolarizability@xcite lead to a more rigorous argument of why the original calculations , which appeared to truncate the sum rules , did not violate the higher - level sum rules.@xcite a more important concern was the use of the three - level ansatz , which states that the fundamental limits can be calculated by assuming that only two excited states contribute .
given that the sum rules allow transition moments between excited states to be large , and potentially many states can contribute , one might expect that the calculations of the fundamental limit would underestimate the upper bounds .
this is clearly not the case .
while a system with only two dominant excited states yields the fundamental limit when the ratio of the oscillator strengths to each of the two excited states is chosen appropriately , systems in which many excited states contribute can also reach the fundamental limit .
the implication is that all of the large contributions from transitions between excited states cancel .
we have shown that when assigning transition moment matrix elements and energies randomly , but under the constraint of the sum rules , values of the intrinsic hyperpolarizability approach unity .
this is in contrast to the intrinsic hyperpolarizability , which when calculated from an optimized potential energy function , falls bellow 0.71 .
the distribution of the values of @xmath64 is found to be well approximated by a cycloid - like function . when the energy - level spacing is weighted in a way that forces the lower energy states to have more closely - spaced energies
, the distribution falls below the cycloid function , suggesting that more sparse energy - level spacing of the lower levels yields a larger intrinsic hyperpolarizability an observation consistent with past analytical studies . as the intrinsic hyperpolarizability increases
, it is found that fewer states contribute even though the transition moments between many states are nonzero and can be large .
furthermore , when @xmath64 is near unity , only three states contribute significantly , and the energy level spacing between the first and second excited states is large compared with the energy difference between the first excited state and the ground state .
this result is consistent with the three - level ansatz .
furthermore , these matrix elements and energies yield an energy function @xmath118 and transition moment function @xmath119 that accurately predicts the hyperpolarizability , which confirms the validity of the type of analysis proposed earlier,@xcite and applied to an amylose helix.@xcite we have also shown that our approach of using a finite number of states , @xmath14 , does not lead to pathologies .
we did so by constructing a state space of dimension @xmath18 , whose transition moments and energies satisfy the higher - order sum rules ; but , lead to the same hyperpolarizability as for the @xmath14-dimensional state space .
since we can construct an infinite - dimensional space by successively adding one state at a time , we conclude that the full set of sum rules are obeyed and the hyperpolarizability remains unchanged from the @xmath14-dimensional one .
we question whether or not the energy levels and transition moments that are used to calculate the hyperpolarizability when @xmath120 are derivable from a simple hamiltonian with a position - dependent potential energy function
. it may very well be the case that , while the sum rules are derived from commutations relationships that assume a potential energy function that depends on position , using only the sum rules as a constraint to determining transition moments and energies may lead to results that belong to a class of more complex hamiltonians that also satisfy the commutation relationship @xmath121,x ] = \hbar^2/m$ ] .
if this is so , additional constraints may be needed to ensure that such systems are physically reasonable . | the off - resonant hyperpolarizability is calculated using the dipole - free sum - over - stats expression from a randomly chosen set of energies and transition dipole moments that are forced to be consistent with the sum rules .
the process is repeated so that the distribution of hyperpolarizabilities can be determined .
we find this distribution to be a cycloid - like function .
in contrast to variational techniques that when applied to the potential energy function yield an intrinsic hyperpolarizability less than 0.71 , our monte carlo method yields values that approach unity .
while many transition dipole moments are large when the calculated hyperpolarizability is near the fundamental limit , only two excited states dominate the hyperpolarizability - consistent with the three - level ansatz . |
as a model for the interface we consider the unduloid , a family of constant mean curvature surfaces that appear in the study of soap films @xcite .
these have also been used to describe the pearling instability in fluid membrane tubes @xcite . for our purposes ,
the unduloid surfaces have the convenient property that their geometry can be tuned continuously and rather simply from cylindrical to spherical .
this family of surfaces is described by two parameters : @xmath0 which varies from @xmath1 for the cylinder to @xmath2 for a necklace of spheres , and the size scale @xmath3 which sets the radius of the cylinder at @xmath4 @xcite .
intermediate values of @xmath0 describe an undulated cylinder surface ( see fig .
[ fig : unduloid ] ) . following ref .
@xcite , the surface can be represented parametrically with a vector that maps @xmath5 , the coordinate along the cylinder axis , and @xmath6 , the azimuthal coordinate , to points on the interface between the charged and hydrophobic blocks , @xmath7 , where @xmath8 and @xmath9 we see that @xmath10 is periodic as @xmath11 .
it is important to note that , while @xmath3 describes the radius of the cylinders , the radii of the spheres at @xmath4 is @xmath12 .
note that at small @xmath13 , the unduloid describes a simple sinusoidal modulation of the cylinder radius , @xmath14 in the case of soap films , unduloids are necessarily unstable due to their finite length .
for diblock copolymers , however , we will see the situation is far more complicated .
the free energy of polyelectroyte copolymer micelles is a sum of four terms , @xmath15 where @xmath16 is the stretching energy of the chains in the core , @xmath17 is the interfacial energy between the core and dissolved brush domains , @xmath18 is the stretching energy of the brush , and @xmath19 is the contribution to the free energy from the charged groups on the chain and the ions in solution . in the remainder of this section we provide results and motivation for the expressions , leaving their detailed evaluation to the appendix .
if the polymer block composing the core region is sufficiently hydrophobic , then the concentration of solvent in the core region is negligible and the chains in the core are in the molten state with constant density @xmath20 .
thus , the techniques of self - consistent strong - segregation theory of diblock copolymers in a melt can be applied @xcite . in the strong - segregation limit ,
the width interface between domains is small compared to the size of the domains , and the free energy can be expressed in terms of the geometry of the interface . to this end
we employ the `` wedge model '' of milner and olmsted @xcite in which we partition the core region contained by the unduloid surface into a series of wedges small enough so that all chains within effectively experience a single , fixed geometry .
the total stretching free energy is computed by , @xmath21 where @xmath22 denotes an integral over the volume of the core domain , appropriately apportioned into infinitesimal wedges , @xmath23 is the distance along the chain trajectory to the interface , @xmath20 is the density of monomers in the core , @xmath24 is the molecular weight of the chains in the core and @xmath25 is the kuhn length .
we assume that the chains meet the interface along the surface normal .
one advantage of using the unduloid representation of the interface rather than a cylinder with a cosine modulation of the radius is that , for all values of @xmath0 , the chain paths do not cross , allowing us to follow the transition all the way from the cylinder morphology at @xmath26 to a chain of spheres at @xmath4 .
we show in the appendix that the stretching energy per chain can be put into the form @xmath27 where @xmath28 is the number of chains , @xmath29 is the dimensionless volume of the micelle core , @xmath30 is the core volume , and @xmath31 is the dimensionless stretching energy for the chains in the core . since this term represents the entropic cost of confining a chain within a brush , the free energy simply scales as @xmath32 , where @xmath33 is the unperturbed radius of gyration of the polymer block . at the interface between the core and dissolved brush domains ,
the hydrophobic block is exposed to solvent , as well as the polyelectrolyte block .
assuming that the interface is sufficiently narrow , we can capture energetic cost of this exposure by introducing a free energy per unit area , or surface tension , @xmath34 .
in general , such a surface tension will have a complicated dependence on the relative concentrations of solvent and polymer at the interface , free charge and salt concentration , _ etc . _ we will treat , @xmath34 , as an unknown phenomenological parameter effectively fixed for a given system at constant temperature . the interfacial energy per chain is then given by @xmath35 where @xmath34 is the surface tension ( in units of @xmath36 ) , @xmath37 is the interface area , and @xmath38 the dimensionless area of the unduloid surface . in the limit that the concentration of free ions ( both counter ions and salt ions ) in the brush is large ,
the primary effect of these free charges is to maintain electroneutrality @xcite . in this limit , the so - called osmotic - brush " regime ,
the stretching of the chains is balanced by the entropy of the counterions confined to the brush , and the counterions locally neutralize the charge on the chains @xcite . in the present calculation
, we consider chains that are equally - stretched @xcite , and neglect polymer trajectory fluctuations .
the latter approximation can be justified in the low - salt limit and the limit of large polyelectrolyte block monomer number , @xmath39 .
fluctuations of chain path around the mean trajectory have characteristic size , @xmath40 , while simple scaling considerations show that in the low - salt limit the brush height grows as @xmath41 @xcite .
thus , in the asymptotic , large @xmath39 limit , such fluctuations can be ignored .
further , jusufi _ et al . _
@xcite have found that these these approximations lead to reasonable quantitative agreement with simulations for star polymers because the osmotic pressure of the counterions lead to highly stretched chains .
however , their model includes two effects which we neglect here : ( 1 ) the condensation of counterions , which we simply absorb into a redefinition of the fraction of charged monomers , @xmath42 , and ( 2 ) the impartial neutralization of the brush due to the escape of counterions to solution .
a schematic of the chain model is shown is figure [ fig : model ] .
we assume a brush height , @xmath43 , and gaussian chains with stretching energy @xmath44 where for simplicity we take the kuhn length of the polyelectrolyte block to be , @xmath25 , as it is for the insoluble block . to allow for appropriate generality
, we introduce into our _ ansatz _ another parameter @xmath45 which describes the average direction of the chains in the brush , @xmath46 where @xmath47 is the unit normal of the interface ( see fig .
[ fig : model ] ) . in the regions of high negative mean curvature ,
the density of chains , and hence the osmotic pressure , is much larger than the rest of the brush .
since charged brushes do not typically overlap @xcite , we can allow the configurations to relax away from overlap in two ways .
one on hand , the chains trajectories can tip radially away from the high density region , as is described by a @xmath48 configuration .
we could further allow the brush to compress by imposing a sinusoidal variation of the brush height , @xmath49 .
for configurations without overlap , we set @xmath50 , and otherwise , we choose the smallest @xmath51 such that the brush does not overlap .
indeed , we find over a large range of parameters the free energy favors uncompressed states , with @xmath52 . by suitably rescaling @xmath18 , the dimensionless corona stretching per chain ,
@xmath53 $ ] , can be defined as @xmath54}{{{\cal v}}(\epsilon)},\ ] ] where @xmath55 is the height of the brush measured in units of the core size .
finally , we assume that each chain has a fraction @xmath42 of charged monomers equally distributed along the chain .
that is , we assume that the charge distribution is quenched . the counterion contribution to the free energy is then @xcite @xmath56,\nonumber\end{aligned}\ ] ] where @xmath57 is the position - dependent concentration of positive and negative ions , @xmath58 is the concentration of ions in bulk solution , and the integral is over the volume of the brush .
the mean - field counterion densities should , properly , come from solving the poisson - boltzmann equation for the charge distribution described by @xmath59 , where @xmath60 is the local concentration of monomer in the brush which depends on the brush geometry ( see appendix ) . to derive this expression ,
we have assumed local electroneutrality , @xmath61 @xcite , which is valid in the osmotic regime @xcite . from the poisson - boltzmann equation
, we also have the relation @xmath62 for a @xmath63 electrolyte .
this allows us to derive expressions for the ion concentrations in terms of @xmath64 and @xmath58 : @xmath65 strictly speaking , we are considering the limit of the poisson - boltmann theory in the limit that the debye screening length , @xmath66 , is small compared to brush dimensions @xcite .
this is valid for the case when the concentration of free charge ( counterions and salt ) in the brush is sufficiently large .
we note that the expression for the osmotic pressure used in a similar mean - field calculation for polyelectrolyte copolymers by borisov and zhulina @xcite can be obtained by taking the @xmath67 limit of the above expressions .
in general , one expects that some fraction of the counterions will be condensed onto the chains we can absorb this into a redefinition of the fraction of charged monomers @xmath42 and assume that @xmath42 does not change overly much when the interface geometry changes .
additionally , some fraction of the counterions will escape the corona , an effect we also ignore in our approximations .
this will lead to region near the edge of the brush which is non - neutral .
however , we expect that the size of such a region will be proportional to the electrostatic screening length , which is small compared to the brush size .
it seems doubtful that these approximations will change the overall qualitative conclusions of this paper , though they are likely to play a quantitative role in some aspects of our calculation .
this will be discussed where appropriate in the following .
given the energetics described here we obtain mean - field results by minimizing the free energy per chain over unconstrained parameters .
for the case of the equilibrium calculations discussed below , we minimize all free parameters .
the counterion free energy density does not depend on the core size @xmath3 , allowing us to minimize over @xmath3 analytically , only on the ratio of brush height to core size , @xmath68 .
we retain three parameters to be minimized over numerically : the brush height @xmath69 , the undulation parameter @xmath0 , and the brush chain direction @xmath45 .
this gives us @xmath70 \}},\ ] ] where @xmath71 .
we computed mean - field phase diagrams for crew - cut " ( @xmath72 ) , symmetric ( @xmath73 ) and hairy " ( @xmath74 ) micelles .
the free - energy minimization was performed for fixed charge fraction , @xmath42 , and salt concentration , @xmath58 .
we assume for simplicity that @xmath75 , and we consider copolymers composed of @xmath76 monomers per chain , roughly equivalent to the experiments of bendejacq _ et al _ @xcite .
we set the surface tension to @xmath77 and find that hairy micelles ( @xmath72 ) are computed to have core radii in the range of 5 to 10 nm ( assuming @xmath78 5 ) , which is in good agreement with the experimental observations of ref .
@xcite . the phase behavior for hairy micelles ( @xmath72 ) as a function of charge fraction @xmath42 and dimensionless salt concentration @xmath79
is shown in fig .
[ fig : phasediagramf0p3 ] . the free energy per chain of cylindrical and spherical micelles are equal along the dashed line . at high salt and low polymer charge cylindrical micelles
are stable , while at low salt and high charge spherical micelles become stable .
however , cylindrical micelles that persist into the stability region of spheres become unstable to undulation .
these undulations are driven by the frustration induced by the counterions , which prefer interfaces with positive gaussian curvature in order to increase their free volume . however
, this tendency toward the formation of spheres must be balanced by the significant increase in free ion concentration in the regions of negative gaussian curvature which is only partially mitigated by the migration of chains toward the positive curvature region ( i.e. surface density of chains is reduced in the saddle regions ; see fig .
[ fig : model ] for a schematic description ) .
therefore we conclude that the undulated cylinder shape is _
metastable_. in figure [ fig : epvsalpha ] we plot @xmath13 and @xmath45 as function of @xmath80 for fixed salt concentration ( @xmath81 ) .
when the cylinders first become unstable at low charge fraction , the undulated cylinders have eccentricities on the order of @xmath82 and @xmath83 so that the chains are almost nearly directed normal to the interface . as the charge fraction
is increased , the increased osmotic pressure within the brush increases its height ( see fig . [
fig : epvsalpha ] ) .
this effect leads to an increase in the monomer and counterion concentration in the negative curvature regions .
the subsequent rise in osmotic pressure leads to a tendency of the chain trajectories to tip toward the radial direction .
once @xmath45 decreases to zero and the brush chains extend radially from the cylinder axis , the counterions no longer gain any entropy from the extra interfacial curvature of the unduloid .
in fact , the core free energies , the core stretching and interfacial energy , are minimal for the @xmath84 cylindrical state . and since the brush free energy is not lower for an unduloid state with @xmath85 than the @xmath84 state , @xmath13 decreases back to 0 for large @xmath80 .
the boundary of the undulated cylinder region at high charge fraction is determined by the condition @xmath26 ( see fig . [
fig : epvsalpha ] ) . the phase behavior of symmetric ( @xmath73 ) and crew - cut ( @xmath74 )
micelles is shown in figs .
[ fig : symphase ] and [ fig : crewphase ] respectively , and is very similar to the hairy case . as we progress toward the crew - cut limit , the cs transition moves toward larger charge fractions and lower salt concentrations .
this may be due to the core energy which prefers cylindrical micelles over spheres and is dominated by the interfactial term , ( [ eq : int ] ) , becomes more important as the size of the core block , @xmath24 , increases . as the core chain length increases , the undulated cylinder behavior no longer disappears at large charge fraction for @xmath86 . as in the case of hairy micelles , however , spheres become energetically favorable well before the uc region in both cases .
given our model for the free energy of micelles constructed from the unduloid surfaces , we build an _ ansatz _ for the shapes of the ends to micelles of cylindrical micelles from consider piecewise constant mean curvature surfaces .
this is done by `` gluing '' two unduloids quarter - periods and a spherical hemisphere to the end of a cylinder with radius @xmath3 ( see fig . [
fig : endansatz ] ) .
one of the unduloid pieces changes radius from @xmath3 to @xmath87 , the neck radius .
the second piece increases the radius from @xmath87 to the spherical cap radius , @xmath88 .
since we assume that the micelles are effectively infinite , the end pieces can increase in volume by pulling chains from the infinite cylindrical portion , and thus , raise or lower the energy of that particular chain relative to its energy in the cylinder .
thus , to compute the free energy of these ends we consider that the ends are in contact with an infinite source of polymer chains at fixed chemical potential , which is set by the free energy per chain of the infinite cylindrical state .
the cylinder radius , @xmath3 , is set to the equilibrium value of an infinite cylinder , and we minimize the free energy of the end over all remaining unconstrained parameters .
many of the features of the characteristic end shapes can be explained by considering the free energy per chain in different geometries . in the region of cylinder stability ( at low charge and high salt ) , chains in the endcap have a higher free energy than those in the bulk cylinder , and therefore preferentially migrate into the cylinder region .
this has the effect of decreasing the radius of the spherical endcap , @xmath88 , giving the appearance of the interface at @xmath89 in fig .
[ fig : endshape ] . in the spherical stability regime ( at high charge and low salt ) , we find that the endcap , @xmath88 , is generically larger than the cylinder radius , @xmath3 . since in this regime the free energy of chains within the spherical bulges is lower than for chains with the cylindrical portion , chains preferentially migrate toward the spherical cap , increasing the radius of end cap , @xmath88 .
to best approximate the preferred spherical geometry , the neck radius @xmath87 , pinches down to well below @xmath3 .
this procedure yields a spherical cap radius @xmath88 very close to the ideal spherical micelle size .
this characteristic bulbous " end shape has been widely observed in a number of polymeric micelle systems @xcite , and is similar to the ends conjectured to appear in the context of bicontinuous phases of emulsions @xcite . however , despite the preference for a spherical micelle configuration indicated by these bulbous end shapes , the free energy always has a minimum at finite neck size .
indeed , the high osmotic pressure induced by the large ion concentration in the saddle regions of the charged brush domain lead to sizable energy barriers to `` pinching off '' ( @xmath90 ) .
these barriers are discussed in detail below .
since undulated cylinders are a metastable state , we consider here the free energy barriers governing the eventual breakup of the micelles into spheres .
diblock copolymer kinetics is a subject of ongoing theoretical @xcite and experimental @xcite research . since diblocks in solvent are the polymeric equivalent of amphiphilic surfactants , it is instructive to compare the kinetics of their respective micellar distributions . in surfactants ,
there is a substantial fraction of free molecules and the primary kinetic mechanism for morphological changes is the aniansson - wall ( aw ) mechanism @xcite .
briefly , the exchange of single surfactant molecules ( unimers ) from micelle to bulk solution is fast enough to allow the direct formation of spheres and the subsequent depletion of molecules from cylindrical to spherical micelles . in the higher molecular weight copolymers , however ,
the concentration of free chains is significantly suppressed @xcite and , consequently , the aw unimer exchange mechanism is drastically slowed @xcite . instead , dormidontova has found that fission - fusion events of micelles likely play a significant role in the relaxation kinetics of copolymers solutions @xcite .
this suggests that the kinetic pathway connecting cylindrical and spherical micelle distributions may be dominated by states of fixed polymer number and intermediate interfacial geometry , such as undulated cylinders @xcite .
we turn our attention to the calculation of the free energy barriers for the fission of a cylindrical ( or undulated ) micelle end into spherical micelles .
our end shape _
ansatz _ allows us to compute the energetic barrier for budding a sphere from the end of a cylindrical micelle by continuously tuning the neck size , @xmath87 .
a schematic free energy plot for this pathway is depicted in fig .
[ fig : budpath ] .
we consider a pathway which is divided into three stages : ( 1 ) contraction of the neck radius , ( 2 ) scission of the neck , and ( 3 ) bud relaxation .
while , in principle , we could imagine a pathway in which @xmath87 contracts continuously to zero ( or at least some molecular size ) , we find that allowing for neck scission lowers the over barrier considerably . therefore , once the neck size has decreased below some critical size , the bud neck breaks , thereby exposing an area of the core @xmath91 to solution . at the point of neck scission , we estimate the cost of exposing core monomers to solution as @xmath34 , the interfacial surface tension , times the exposed surface area . to compute the minimum energy pathway within our _ ansatz _
we minimize the free energy barrier , @xmath92 , over this critical neck radius .
our calculation for the barrier height for end fission of spherical micelle from a cylinder end is plotted in figure [ fig : barrier ] for hairy micelles ( @xmath72 ) and for low salt ( @xmath93 ) .
we find that @xmath92 _ decreases _ for increased @xmath42 .
when @xmath42 is small , the barrier heights can be quite large ( thousands of @xmath36 ) . for larger
@xmath42 , however , the barrier heights fall precipitously to tens or hundreds of @xmath36 . in comparison to the case neutral copolymer melts @xcite , the free energy barriers computed here are quite large , up to 1000 times larger . a simple consideration of the free energy scale set by the brush free energy shows that the excluded volume interaction induced by the brush electrostatics should indeed lead to free energy barriers of this magnitude .
while excluded volume effects are screened in molten polymer brushes , the cs transition is frustrated by the increased osmotic pressure in the neck region of the end bud .
typically , such an end bud will contain @xmath94 polymer chains , and each of those chains entrains @xmath95 counter ions within the dissolved brush .
given that each of those counterions represents roughly @xmath36 worth of free energy due to the entropy lost in confinement , it is then not surprising that such a topological transition could cost thousands of @xmath36 .
nonetheless , we see from figure [ fig : barrier ] that changing the effective charge of the polyelectrolyte block can _ decrease _ the barrier at least two orders of magnitude .
the qualitative explanation of this effect is simple . as @xmath80 increases
, the equilibrium size of the micelles decreases , indicating that fewer chains comprise the micelle ends , leading to an overall decrease in @xmath92 .
the barrier heights for hairy micelles are plotted on a log - log scale in figure [ fig : logbarrier ] for a number of different salt concentrations .
we find two characteristic regimes of power law behavior , one at low charging and one at high charging . at high salt concentration and large @xmath42
, we find that the barrier height decreases as @xmath96 . in this regime ,
the decrease in neck radius @xmath87 is small and the free energy barrier is dominated by direct scission . here
, the dependence on @xmath42 comes directly from the equlilibrium scaling of the cylinder radius with @xmath80 at high salt in the limit @xmath97 , @xmath98 @xcite .
largely , we expect the equilibrium neck radius to scale with @xmath3 , therefore , direct scission predicts a free energy barrier , @xmath99 , which yields the observed power law .
we have developed a free energy expression for diblock copolymer cylinders with a hydrophobic block and a charged corona , and used this to understand the stability of various axisymmetric structures .
we use a family of constant mean curvature surfaces called unduloids to parametrize possible interfaces from cylinder to a necklace of spheres . this _
ansatz _ allows us to probe the full , non - linear dependence of the free energy on the micelle surface geometry .
our results indicate that there is a metastable undulated cylinder phase , as observed in experiments @xcite which occurs generically in parameter ranges where the free energy per chain for spherical micelles is lower still .
we have studied the shape of the spherical endcaps and found , in particular , that in the regime in which cylinders are stable , the ends show a distinct decrease in radius below the equilibrium value . when spherical micelles are preferred in equilibrium , the spherical cap radius is larger than the cylinder radius and is close to the preferred spherical micelle size .
we use these results to compute dependence the free energy barriers for budding spherical micelles from cylindrical micelles on polymer charge , @xmath80 , and solution salt concentration , @xmath58 .
our calculation of the free energy is variational and therefore our values for the free energy barrier to budding , @xmath92 , necessarily provide upper bounds . however , certain conclusions from this study are robust . indeed , these results along with the fact that the aw step - wise kinetic mechanism is suppressed in polymeric systems @xcite strongly support the suggestion that distributions of polymer micelles in selective solvent non - ergodic , " with micelles essentially trapped within a fixed , quenched topology @xcite .
in contrast with neutral copolymer solutions , however , we find here a strong , tunable dependence of barrier height on polymer charge and salt concentration .
kramers kinetic theory predicts that the cs transition rate , @xmath100 should depend on the the free energy as @xmath101 @xcite . at small @xmath42 , the cs pathway investigated here
is kinetically arrested by large barrier heights . supposing the aw unimer exchange mechanism is also sufficiently slow , these large barrier heights suggest that the metastable undulated cylinder regime may be observed .
such undulated cylinders have been observed experimentally in a number of amphiphilic polymeric systems , both charged and uncharged @xcite . at large @xmath80 and low @xmath58
, @xmath92 may fall to sufficiently low values that the transition to spherical micelles may proceed on experimental time scales .
this suggests that the observed `` phase boundary '' between cylindrical and spherical micelles in ref .
@xcite , may reflect that fact that these kinetic barriers depend quite strongly on @xmath80 and @xmath58 .
given that @xmath92 varies by two orders of magnitude , we expect an extremely sharp dependence of the @xmath100 on @xmath80 and @xmath58 over the same parameter range .
the geometry of self - assembled structures is often cast in terms of constant mean curvature ( cmc ) surfaces @xcite .
more generally , the energy of any interface can be expanded in powers of the curvature , resulting in the helfrich energy , @xmath102 where @xmath103 is the total micelle interfacial area , @xmath104 is the mean curvature , @xmath105 is the gaussian curvature , @xmath106 is the spontaneous curvature , which comes from the molecular asymmetry of the chains and the corona - solvent interaction .
the parameter @xmath34 is the surface tension .
finally , @xmath107 and @xmath108 are effective bending rigidities for the mean and gaussian curvatures respectively , which must be derived from a microscopic model . for @xmath109 ,
this energy describes the shape of soap bubbles , and cmc surfaces minimize the energy in this limit . in the context of diblock copolymers
@xcite or fluid membrane shapes @xcite , the additional curvature terms modify the shape and stability of various structures .
higher order curvature terms are typically neglected as being subleading , but we will show equation ( [ eq : helfrich ] ) fails to describe our results , even qualitatively , without them . we restrict ourselves to interfaces belonging to the unduloid family of surfaces capped at both ends .
the energy reduces to @xmath110 a_{tot}(\epsilon ) - 4 \pi \bar{\kappa},\ ] ] where we note that a cylinder with two ends has @xmath111 from the gauss - bonnet theorem . if we choose the length of the cylinder to conserve the volume , the total area is given by @xmath112 v_{tot}$ ] , where @xmath113 and @xmath114 are the reduced area and volume of a half undulation wavelength ( see appendix ) and @xmath115 is the volume of the cylinder .
the only dependence on @xmath0 comes from the volume and area of the interface ( recall that @xmath116 does not change with @xmath0 ) which , at fixed @xmath3 , is minimum for @xmath117 no matter the value of spontaneous curvature .
this reflects the fact that spheres minimize the area for a fixed volume , and indeed , this is a manifestation of the well - known rayleigh instability of fluid interfaces @xcite .
the spontaneous curvature only sets the overall size scale of the micelles , not their topology .
we conclude that higher order curvature terms to equation ( [ eq : helfrich ] ) are required to understand the local stability of undulated cylinders in terms of universal geometrical measures of the block interface @xcite .
this is , perhaps , not surprising since higher order corrections introduce length scales on the order of @xmath3 for axisymmetric shapes @xcite . in the case of vesicles , where equation ( [ eq : helfrich ] ) is known to apply ,
these corrections are controlled by the membrane thickness @xcite .
we leave the development of a geometrical shape equation for diblock copolymer micelles to future work .
the authors are pleased to acknowledge discussions with d. bendejacq , v. ponsinet , p. pincus , and r. kamien .
gmg was supported by nsf grant dmr04 - 04507 .
cds was supported by nsf grants dmr01 - 29804 , dmr05 - 47230 , and the pennsylvania nanotechnology institute .
in addition , both authors were supported under nsf grant dmr05 - 20020 ( mrsec ) , acs petroleum research fund , and a generous gift from l.j .
in this appendix we present details concerning the geometry of unduloid surfaces which will be useful in our free energy calculation .
given the parametric representation of @xmath10 by eqs .
( [ eq : z ] ) and ( [ eq : r ] ) we have the differentials , @xmath118 and @xmath119 the tangent vectors for the two principle directions of the surface are given by , @xmath120 and @xmath121 \ .\ ] ] from these we compute the differential element of area , @xmath122 and the surface normal , @xmath123}{\sqrt{1+\epsilon^2 + 2 \epsilon \sin u } } \ .\ ] ] since the surface is periodic we concern ourselves only with a half period of the undulation , @xmath124 $ ] .
the volume per half period can be computed , @xmath125 the dimensionless volume , @xmath114 , is then defined by @xmath126 . finally , we surface area of a half period is computed by , @xmath127 and the dimensionless area , @xmath113 , is defined by @xmath128 .
in this section we compute the core stretching ( [ eq : cst ] ) . as described in sec .
[ sec : model ] , we divide the core region into infinitestimal wedges , which contain polymer chains extending in the same direction .
we assume that the core chains ( and wedges ) extend along the surface normal .
as shown in figure [ fig : map ] the vector , @xmath129 , describes the extension of the wedge . from the normal stretching condition
, we have the mapping from a point on the surface to a point along the @xmath130 axis , located a @xmath131 ( see figure [ fig : map ] ) .
this mapping is given by , @xmath132 in general , this mapping must be consistent with the constraint the chain paths ( and wedges ) do not cross .
this means that the mapping must be one - to - one .
this constraint is satisfied for @xmath133 .
this yields an inequality which must be satisfied by surfaces we consider , @xmath134 \leq 1\ .\ ] ] it is not difficult to show that all unduloid surfaces satisfy this inequality . to compute the stretching integral
defined in eq .
( [ eq : cst ] ) , we must compute the volume distribution within the wedge , which we do by considering the dimensions of the wedge in the @xmath135 and @xmath136directions as a function of distance to the @xmath130 axis , @xmath137 .
we will denote these directions by @xmath138 and @xmath139 , respectively ( see figure [ fig : core ] ) .
these are given by , @xmath140 and @xmath141 where @xmath142 . from this
we compute the area distribution in the core wedges by @xmath143 . with this
the volume per core wedge can be computed , @xmath144 and the surface density , @xmath145 , of chains reaching the interface at , @xmath146 we plot @xmath145 for various unduloid shapes in figure [ fig : sigma ] . for cylinders ( @xmath84 ) and spheres ( @xmath147 )
the surface density is constant , while for intermediate shapes @xmath145 drops in the saddle regions .
finally we compute the stretching free energy for a half period of the core region , @xmath148 where the dimensionless stretching is defined on the last line .
dividing this with the number of chains in a half period of core is @xmath149 , we obtain eq .
( [ eq : cst ] ) .
in figure [ fig : brushtrajectory ] we show the geometry of the chain trajectories in the outer polyelectrolyte brush .
the vector @xmath150 gives the extension of brush chains from the interface surface at @xmath10 to the surface described by the tips of this outer brush at @xmath151 .
as described in sec .
[ sec : model ] make the assumption that all chains are uniformly stretched , with their free ends constrained to tips of the brush .
given this assumption , it is straightforward to compute the free energy cost of extending the brush within a half period of undulation , @xmath152 \ .\end{aligned}\ ] ] dividing by the number of chains in the core , we obtain eq .
( [ eq : bst ] ) .
next we compute the brush monomer concentration , @xmath153 , which depends on the brush geometry .
therefore , we need to determine the volume distribution as we did for the core brush . in particular , we divide the outer brush into wedges extending along the mean polymer trajectory and we compute the area distribution within these wedges as we did in the case of the core ( see fig .
[ fig : brushtrajectory ] ) . here , we define @xmath154 as we did before , and @xmath155 is the dimension of the wedge associated with the direction perpendicular to @xmath156 which we denote @xmath157 .
these can readily be computed from the geometry , @xmath158 and @xmath159 we can compute the area available to polymers in the wedge at a distance , @xmath160 , from the interface by @xmath161 .
if a given wedge leaves the surface from @xmath10 , then it contains @xmath162 polymer chains .
given that we consider the corona chains to be evenly stretched throughout the brush , a chain leaves , @xmath163 monomers between @xmath160 and @xmath164 , in a differential volume , @xmath165 .
therefore , the position dependent monomer concentration is , @xmath166 we use this relation in eq .
( [ eq : charge ] ) to compute the concentrations of free ions in the brush . | we study the cylinder to sphere morphological transition of diblock copolymers in aqueous solution with a hydrophobic block and a charged block . we find a metastable undulated cylinder configuration for a range of charge and salt concentrations which , nevertheless , occurs above the threshold where spheres are thermodynamically favorable . by modeling the shape of the cylinder ends
, we find that the free energy barrier for the transition from cylinders to spheres is quite large and that this barrier falls significantly in the limit of high polymer charge and low solution salinity .
this suggests that observed undulated cylinder phases are kinetically trapped structures .
diblock copolymers , formed by joining two immiscible polymers at one end , are a model system to study many aspects of self - assembly and order - order transitions from one self - assembled state to another @xcite .
self - assembly in diblock copolymers is driven by two competing effects : the immiscibility of the two blocks drives their phase separation , yet the connectivity of the blocks prevents a true macroscopic phase separation . even in the simplest case of copolymer melts , this competition leads to the formation of a variety of generic structures , including spherical and cylindrical micelles , lamellae and bicontinuous gyroid phase @xcite .
the selection of these various morphologies is determined by a balance of surface tension between microphase separated regions and the chain stretching of the polymers and can , in certain cases , be analyzed purely from the point of view of the interfacial geometry @xcite . in selective solvents , on the other hand ,
the morphology of copolymer structures is also affected by solvent - polymer interactions , widening the both the range of possible structures and the ability to control morphological transitions by tuning the solvent conditions . moreover , in solution , the volume - filling constraints of the molten state are relaxed . in this way ,
diblock copolymers are the analogues of surfactants , and exhibit many of the typical self - assembled morphologies : spherical micelles , wormlike ( cylindrical ) micelles , and lamellae @xcite . despite the similarities , however
, these systems also display a variety of novel structures , including wormlike micelle networks @xcite , large compound vesicles @xcite , and undulated cylinders @xcite . in this paper
, we study the undulated cylindrical morphology by considering axisymmetric , diblock copolymer micelles dispersed in aqueous solution , concentrating on the case of copolymers composed of a hydrophobic block and a polyelectrolyte block . in part
, this focus is due to the fact the self - assembly in polyelectrolyte copolymer solutions has been well characterized experimentally @xcite . while similar experiments have address the morphological transition between wormlike micelles and spheres for neutral block copolymers @xcite , the added control provided by modulating the charge and salt conditions of the polyelectrolyte domains allow for a systematic investigation of , in particular , the cylinder to sphere ( cs ) transition .
bendejacq _
et al . _
@xcite have performed such experiments at controlled salinity and ionic strength and find that when the polyelectrolyte domains are highly charged , the cs transition proceeds by the undulation of the cylindrical micelles followed by the eventual breakup into spheres . at certain intermediate ionic strengths and salt concentrations , however , undulated cylinders were also observed to have very long lifetimes , and were suggested as a candidate for a new intermediate phase between cylinders and spheres in ps - paa diblock copolymers .
we develop a mean - field model for diblock copolymer micelles with interfacial geometry between straight cylinders and individual spheres .
experimentally , this can be realized by first forming the cylinders , then changing solvent conditions to frustrate the geometry toward the formation of spherical micelles , as was the method in the experiments of ref .
@xcite . to this end
, we construct a class of micellar morphologies whose geometry can be continuously varied from cylindrical to spherical and compute the free energy within a variational , mean - field scheme .
our main result is that under certain salt and charge conditions , cylindrical micelles become unstable to the formation of undulated cylinders .
these undulated cylinders are metastable , however , and their observation by bendejacq _
et al . _
@xcite must indicate a large kinetic barrier to the breakup of undulated cylinders to spheres .
our model allows us to calculate an upper bound on the free - energy barrier for fission of a cylinder into spheres , and we find a strong dependence of the barrier height on salt concentration of the solution and the effective charge of the polyelectrolyte blocks .
combined these calculations enable us qualitatively to understand the `` phase diagram '' observed in reference @xcite .
we also discuss results on the shape of the micelle ends .
our approach is to construct the simplest possible model for the energetics of undulated cylinders that contains the essential physics . in this spirit
, we will ignore the fluctuations of the chains around their primitive paths and assume the chains in the corona are equally stretched . strictly speaking
, this analysis provides only an upper - bound on the true free energy .
nevertheless , similar calculations are known to be very useful in understanding the equilibrium morphologies of diblock copolymer melts @xcite .
moreover , for _ charged _ polymers in particular , which tend to be strongly - stretched , these techniques have been developed extensively by borisov , zhulina and collaborators @xcite .
recently , jusufi _ et al .
_ @xcite have performed extensive simulations and detailed analytical analysis on charged star polymers and found excellent agreement between simulations and a mean - field electrostatic theory .
the paper is organized as follows : in section [ sec : model ] we discuss the details of our model for the diblock copolymers . in section [ sec : results ] , we present phase diagrams and results on a metastable undulated cylinder phase .
we also discuss the shape of the cylinder ends and compute estimates for the free energy barriers . in section [ sec : discussion ] , we discuss our results . |
this research is supported by the the chinese postdoctoral science foundation and cas k.c .
wong postdoctoral research award fund .
we are grateful to prof .
c.s . huang and prof .
x.m zhang for useful discussions .
u. chattopadhyay , a. corille and p. nath , hep - ph/0202275;m . knecht and a. nyffeler , hep - ph/0111058
; m. knecht , a. nyffeler , m. perrottet and e. de pafael , phys .
lett . * 88 * , 071802 ( 2002 ) ; m. hayakawa and t. kinsoshita , hep - ph/0112102 ; i. blokland , a. czarnecki and k. melnikov , phys .
* 88 * , 071803 ( 2002 ) ; j. bijnens , e. pallante and j. prades , hep - ph/0112255 .
abrams et al . , mark ii collab . , phys .
* 63 * ( 1989 ) 2173 ; b. advera et al .
, l3 collab .
b * 231 * ( 1989 ) 509;i .
decamp et al . , opal collab . , _ ibid .
_ , * 231 * ( 1989 ) 519 ; m.z .
akrawy et al . , delphi collab . , _ ibid .
_ , * 231 * ( 1989 ) 539 . | we investigate the newly observed muon @xmath0 anomaly in the framework of a sequential fourth generation model with a heavy fourth neutrino , @xmath1 . we find that @xmath0 can exclude most values of @xmath2 and put a very stringent constraint on the existence of the fourth generation .
we also obtain bounds of a @xmath3 leptonic mixing matrix elements , @xmath4 if it really exists .
= -30 mm at present , standard model ( sm ) has to face the experimental difficulties which are all relate to leptons .
it seems to indicate the presence of new physics just round the corner will be in the leptonic part .
firstly , there are convincing evidences that neutrinos are massive and oscillate in flavor @xcite .
secondly , the recent measurement of the muon anomalous magnetic moment by the experiment e821 @xcite at brookhaven national laboratory disagrees with the sm expectations at more than 2.6@xmath5 level . defined as @xmath6 ,
the recent measurement of @xmath7 is @xmath8 where the sm prediction is @xmath9 thus , one finds @xmath10 which gives the old @xmath11 deviation .
but the revised difference between experiment and sm is @xmath12 which is now only a @xmath13 deviation @xcite .
although the deviation drops from @xmath11 to @xmath13 , it might seem that the absolute magnitude of the deviation may be a hint of new physics .
there have been a lot of scenarios of new physics proposed to interpret the non - vanishing and positive value of @xmath14 . in this note
, we consider the sequential fourth generation standard model ( sm4 ) to investigate its contributions to @xmath7 . from the point of phenomenology
, there is a realistic question what are numbers of the fermions generation or weather there are other additional quarks or leptons . the present experiments
tell us there are only three generation fermions with @xmath15 neutrinos which mass are less smaller than @xmath16@xcite .
but the experiments do nt exclude the existence of other additional generation , such as the fourth generation , with a @xmath17 neutrino , i.e. @xmath18@xcite . many refs .
have studied the sm4 @xcite , which is added an up - like quark @xmath19 , a down - like quark @xmath20 , a lepton @xmath21 , and a heavy neutrino @xmath22 in the sm .
the properties of these new fermions are all the same as their corresponding counterparts of other three generations except their masses and ckm mixing , see tab.1 , .the elementary particle spectrum of sm4 [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] if there exists a very heavy fourth neutrino @xmath1 , it can contribute to @xmath7 through diagram of fig .
1 . this is a electroweak interaction .
similar to that of quarks , the corresponding lagragian is @xmath23 where @xmath24 , @xmath25 is the ( 2,4 ) element of the four - generation ckm matrix ( @xmath3 ) , @xmath26 reverting back to the diagrams of fig . 1
, we see that the fourth neutrino contribution to @xmath7 is @xmath27 where @xmath28 , @xmath29 is the fine construction constant and @xmath30 we suppose that the 1.6@xmath5 discrepancy of muon anomalous magnetic moment , @xmath31 , is induced by the fourth sequential neutrino @xmath1
. we can use the above equation to get parameter space of @xmath32 and @xmath25 to @xmath2 ( see fig . 2 and 3 ) ,
@xmath33 from fig .
2 , we can see that @xmath32 is taken negative values except for a very narrow range of @xmath2 which is from 58@xmath34 to 80@xmath34 .
in other words , the sign of @xmath35 is only related to @xmath36 mass . only in this narrow mass range ,
@xmath1 gives positive contribution to @xmath0 .
the low bound of @xmath37 we get from @xmath0 is consistent with the present experiments @xcite .
but the upper bound , @xmath38 , seems to conflict with the current experiments statue which there is no any new physics signals upper to several gevs .
the fourth generation particles seems not to be so light .
moreover , from fig .
3 , if we consider the unitarity of the matrix @xmath39 and tiny values of its elements , the reasonable value range of @xmath40 will be more narrow . in summary , we calculate the contribution of the fourth generation to @xmath41 and get an interesting result which we can exclude most values of @xmath2 . considering the new revised deviation of @xmath41 , we give the parameter space of @xmath37 and lepton mixing matrix element @xmath42 .
we find that @xmath41 can constrain on the neutrino mass of the fourth generation : i.e. its mass should be heavier than 58 gev and lighter than 80 gev .
it seems that from the lepton part , the current experiments can impose a stringent constraint on the existence of the fourth generation . |
symmetry has played a fundamental role in the construction of field theories purporting to describe fundamental physics . nowhere is this more true than in the construction of the minimal supersymmetric standard model where symmetries mixing bosonic and fermionic states lead to tightly constrained theories often exhibiting remarkable properties .
this is particularly true for @xmath0 extended supersymmetric yang - mills theories @xcite , which are known to be finite . in this note
we construct field theories which exhibit @xmath0 extended supersymmetry and lorentz violation .
to do so we combine ideas of berger and kosteleck @xcite involving supersymmetric scalar theories exhibiting lorentz violation , and well - known constructions of extended supersymmetric theories involving dimensional reduction ( nicely described in the work of brink , schwartz and scherk@xcite ) .
we begin by establishing notation in the context of the standard construction .
consider a gauge theory involving a single fermion @xmath1 and lagrangian = - f^2 + | , where @xmath2 is the field strength , @xmath3/ig$ ] and @xmath4 is the covariant derivative , d^= ^+ig a^. to simplify notation , we will first consider the abelian case . to implement supersymmetry
we introduce a supercharge , @xmath5 satisfying = 0 , \{q,|q } = 2^p _ , where @xmath6 are the standard dirac matrices \
{ ^,^ } = 2 g^ , and the energy momentum 4-vector @xmath7 generates spacetime translations . the construction of a supercharge is elegantly carried out in the context of _ superspace . _
more precisely , introduce four independent anticommuting variables , @xmath8 and consider the general vector superfield @xmath9 + \frac{1}{2}(\bar{\theta}\theta)^2(d(x ) -\frac{1}{2}\partial_\mu\partial^\mu c(x))\end{aligned}\ ] ] and the @xmath10 operator q= -i _
| -^_. fixing an arbitrary spinor @xmath11 standard analysis@xcite of the operator @xmath12 leads to the supersymmetry transformations defining an @xmath13 supersymmetric theory .
following berger and kosteleck@xcite , we introduce lorentz violation by defining a twisted derivative : @xmath14 where @xmath15 is a symmetric , traceless , dimensionless tensor parametrizing lorentz violation . to obtain a gauge invariant theory
, we also twist the underlying connection : ^= a^+ k^ a_. these perturbations lead to a general vector superfield @xmath16 + \frac{1}{2}(\bar{\theta}\theta)^2(d(x ) -\frac{1}{2}\tilde{\partial}_\mu\tilde{\partial}^\mu c(x))\end{aligned}\ ] ] and perturbed @xmath10 operator = -i _ | -^_. using wess - zumino gauge we obtain a lagrangian @xmath17 where @xmath4 is an auxiliary chiral field and @xmath18 is the twisted field strength , @xmath19 the twisted field strength can be written in terms of the standard sme parameters : @xmath20 , where @xmath21 direct calculation confirms that the action is invariant under the supersymmetry transformations @xmath22 this defines an @xmath13 supersymmetric theory with lorentz violation .
to build an @xmath0 supersymmetric theory we work in @xmath23-dimensional spacetime .
we represent the @xmath24 gamma matrices via @xmath25 where @xmath26 is the @xmath27 identity matrix and @xmath28 and @xmath29 where ^ij = _ 5 ( cc 0 & ^ij + _ ij & 0 ) and the @xmath30 matrices @xmath31 are defined by @xmath32 we consider the lagrangian = - ^2 + k_{\mu\nu}\partial^\nu ) $ ] is the twisted derivative with @xmath33 parametrizing @xmath34 violation and @xmath18 is the corresponding perturbed field strength . imposing both the weyl and the majorana condition and compactifying
, the fermion @xmath1 satisfies @xmath35 where @xmath36 and @xmath37 denote left and right projection operators , respectively , the spinor @xmath38 transforms as a 4 of @xmath39 and the ( independent ) spinor @xmath40 transforms as a @xmath41 of @xmath42 choosing the lorentz violating parameters with care leads to lorentz violating extended supersymmetric theories which are easy to describe .
for example , taking the @xmath33 to vanish in the compactified directions leads to a lagrangian of the form = - ^2 + i | l+ _ _ ij^^ij , where the complex scalar fields @xmath43 transform as a 6 of @xmath39 and are given by @xmath44 the associated action is invariant under the supersymmetry transformations @xmath45 similarly , choosing the @xmath46 parameters to vanish in the spacetime directions @xmath47 we obtain a lagrangian of the form = - f^2 + i | l+ _ _ ij^^ij , where @xmath48 with the matrix @xmath49 containing the effect of lorentz violation in the compactified directions .
the associated action is invariant under the supersymmetry transformations @xmath50 note that if the scalars @xmath51 are identified with physical scalars @xmath43 we restore @xmath39 symmetry and remove any lorentz violating effects .
if , however , the @xmath43 couple to other sectors , lorentz effects may reappear in these sectors .
the above results warrant a number of additional comments : * these constructions can be carried out in the nonabelian case where they yield supersymmetric theories which exhibit lorentz violation@xcite .
* the same techniques can be applied to obtain an @xmath52 supersymmetric theory with lorentz violation .
the construction proceeds by working in 4 + 2-dimensional spacetime and using dimensional reduction@xcite .
* because these constructions involve changing the structure of the underlying superalgebra,@xcite the no - go results of nibbelink and pospelov@xcite do not apply . | we construct a collection of lorentz violating yang - mills theories exhibiting supersymmetry . |
the distinct feature of highly frustrated magnetic materials is a peculiar spatial arrangement of the magnetic ions .
antiferromagnets on typical geometrically frustrated structures , like , for instance , kagome , garnet , and pyrochlore lattices , have an infinite number of classical ground states .
this macroscopic degeneracy precludes any type of conventional magnetic ordering . as a result ,
frustrated magnets remain in a disordered cooperative paramagnetic ground state at temperatures well below the paramagnetic curie - weiss constant @xmath2.@xcite weak residual interactions or quantum and thermal fluctuations usually induce some kind of ordered or spin - glass state at @xmath3 .
a number of geometrically frustrated magnets have been experimentally studied in the past decade.@xcite magnetic pyrochlore compound @xmath0 is one of the prototype examples .
the gd@xmath4 ions have spin @xmath5 and zero orbital momentum , which yields a good realization of a nearest - neighbor heisenberg exchange antiferromagnet on a pyrochlore lattice .
recent specific heat , susceptibility and neutron scattering measurements @xcite have shown that @xmath0 remains disordered over a wide temperature interval below its @xmath6 k. the transition to an ordered phase is presumably driven by weak dipole - dipole interactions and occurs only at @xmath7 k. infinite degeneracy of the magnetic ground state of a frustrated magnet implies the presence of a macroscopic number of local zero - energy modes .
such soft modes correspond to rotational degrees of freedom of a finite number of spins with no change in the total exchange energy . in a pyrochlore structure ( a matrix of corner - sharing tetrahedra on the fcc lattice ) , these groups of spins are hexagons formed by the edges of neighboring tetrahedra , which lie in the kagome planes [ ( 111 ) and equivalent planes ] ( see _ e.g. _ ref . ) . in zero
applied field , if the six spins are arranged antiferromagnetically around one hexagon , they effectively decouple from the other spins and can rotate by an arbitrary angle .
the low - energy hexagon modes have been observed in quasielastic neutron scattering studies on spinel compound zncr@xmath8o@xmath9 , which is a spin-3/2 heisenberg antiferromagnet on a pyrochlore lattice.@xcite the thermodynamic consequence of these local excitations is that a considerable fraction of the magnetic entropy is not frozen down to temperatures much less than @xmath2 .
an interesting effect related to a field evolution of the zero - energy local modes is an enhanced magnetocaloric effect near the saturation field @xmath10 predicted in ref . .
transformation between a nondegenerate fully polarized spin state above @xmath10 and an infinitely degenerate magnetic state below @xmath10 occurs via condensation of a macroscopic number of local modes and produces large changes of entropy in magnetic field .
quantum fluctuations do not destroy such an effect.@xcite in particular , at @xmath11 geometrically frustrated magnets on pyrochlore and kagome lattices have a finite macroscopic entropy , which does not depend on the value of the spin of the magnetic ions .
the finite entropy of kagome antiferromagnet has been calculated exactly,@xcite whereas the corresponding quantity for a pyrochlore lattice remains unknown .
therefore , an experimental observation of the magnetocaloric effect in @xmath0 should be important both from a fundamental point of view , as a physical probe of the local zero - energy excitations in the frustrated ground state below @xmath10 , and because of the possible technological applications .
the enhanced magnetic cooling power of gadolinium gallium garnet @xmath1 has been known for a long time , @xcite although without reference to the frustrated nature of a garnet ( hyper - kagome ) lattice .
experiments have been also performed on other garnets.@xcite thus , a comparison between the two typical frustrated magnets is important since a stronger frustration on a pyrochlore lattice produces the maximal cooling rate among all known types of geometrically frustrated magnets.@xcite in the present work we study experimentally and theoretically the magnetocaloric effect in @xmath0 .
demagnetization of a @xmath0 sample under quasi - adiabatic conditions is performed and compared to a similar study of @xmath1 .
classical monte carlo simulations of an ideal adiabatic demagnetization process are performed for both materials .
entropy variations and the cooling power of @xmath0 in an applied magnetic field are calculated from a combination of the adiabatic demagnetization results and specific heat measurements .
a single - crystal sample of @xmath0 was grown by the method described in ref . and is approximately @xmath12 mm@xmath13 in size ( 32.5 mg by mass ) . for preliminary estimation
, we present the results of specific heat measurements of @xmath0 performed in a quantum design ppms calorimeter at zero field and at @xmath14 koe ( see fig .
[ fig1 ] , upper panel ) .
the curve at @xmath15 was taken from our previous work in ref .
while the high field data were measured in the temperature range from 1.5 to 20 k. instead of two sharp peaks on a zero field @xmath16 dependence at @xmath17 k and @xmath18 k corresponding to the ordering phase transitions , the high field curve demonstrates a broad maximum around 56 k. the lattice contribution to the specific heat extracted by fitting the zero field data at temperatures 3070 k in the debye approximation is shown in the figure by a dashed line .
the temperature dependence of the magnetic entropy ( total entropy with the phonon part subtracted ) obtained by integrating @xmath19 curves are presented at the lower panel of fig .
1 . they give a general overview of the magnetocaloric effect in this compound .
an adiabatic demagnetization of the system down to zero field starting at @xmath14 koe and @xmath20 k ( around @xmath2 ) results in decreasing temperature to 2 k , while the isothermal process at @xmath21 k is accompanied by an entropy increase by approximately one half of the total magnetic entropy .
single crystal sample : @xmath22 zero field , @xmath23 @xmath14 koe ; solid lines are the results of @xmath19 integration , dashed lines represent the phonon contribution . ] in the main part of the experiment , the magnetocaloric effect in @xmath0 is studied in detail by measuring the temperature of the sample in a quasi - adiabatic regime as a function of time and magnetic field . for this purpose , a commercially available thin - film ruo@xmath8 resistor with the resistance calibrated down to 100 mk in fields up to 120
koe was glued onto the sample .
it was also used as a heater to regulate the starting temperature of the experiment .
the sample was suspended on four thin constantan wires ( @xmath24 m in diameter and 5 cm in length ) soldered to the thermometer to make a 4-wire resistance measurement .
the experimental cell was put in a vacuum can immersed in a helium bath held at 1.84.2 k. the heat exchange gas inside the can was absorbed by a charcoal cryopump to a pressure of @xmath25 torr .
the cryopump was equipped with a heater to desorb some exchange gas when necessary to cool the sample during the experiment .
magnetic fields up to 120 koe were generated by a superconducting magnet .
the field @xmath26 was applied perpendicular to the @xmath27 axis .
no correction for the demagnetization factor was made .
the measurements were done at a sweep rate of 10 koe / min . for various starting temperatures ;
dotted lines represent the corrections for the lattice heat capacity ( see discussion ) .
solid lines are obtained by monte carlo simulations with the exchange constant @xmath28 k. ] the temperature versus magnetic field curves for different starting temperatures are shown in fig .
the temperature drop on demagnetization @xmath29 depends on the starting temperature .
it has a maximum around @xmath30 k where the temperature drops by a factor of 10 .
a characteristic feature for all the @xmath31 curves starting below 10 k is that a great deal of cooling occurs in the field range from 120 to 60 koe , which contrasts sharply with a continuous adiabatic cooling ( @xmath32 ) of an ideal paramagnet .
additionally , a parasitic temperature drift @xmath33 ( due to radiation , wiring and the measuring current ) was investigated at several fixed values of the field for temperatures along each curve @xmath31 .
the values of @xmath33 have a smooth temperature dependence within @xmath34 mk / sec changing sign at about 10 k. the experimental data were corrected at each moment of time @xmath35 ( corresponding to each value of the magnetic field @xmath26 ) using a set of measured @xmath33 values by formula @xmath36 where @xmath37 is the raw experimental value , @xmath38 is the corrected temperature .
one should mention , that the above correction procedure is applicable only if the parasitic heating does not strongly affect the whole demagnetization process . in this case , after the correction is applied to the raw data , the process appeared to be fully reversible by the sweep direction , so that the curves obtained on increasing the magnetic field do not differ from those shown in fig .
a more general heat - balance equation @xmath39 ( @xmath40 is a parasitic heat leak ( @xmath41 ) , @xmath42 is the specific heat at a constant field ) leads to formula ( [ tcor ] ) if the function @xmath43 does not change significantly between @xmath44 and @xmath38 .
such an assumption is valid for all the scans shown in fig .
[ fig2 ] except the one starting at @xmath45 k and the corresponding @xmath31 curves are given with the appropriate subtraction .
for the lowest scan , the correction appears to be of the same order of magnitude with the experimental temperature .
in addition , the derivative @xmath46 strongly decreases in the vicinity of the ordering transition which leads to an overcorrection of the experimental data .
this overcorrection is illustrated in fig .
[ fig3 ] by the dashed line .
thus , an ideal adiabatic dependence @xmath31 should lie between the raw experimental data and the corrected curve .
the minimum temperature reached experimentally on demagnetization from @xmath45 k at @xmath47 koe is @xmath48 k. this cooling limit is associated with the magnetic entropy freezing at the transition into an ordered state . when the field is further decreased , a weak temperature increase is observed .
two temperature plateaus ( shown by the arrows in fig . [ fig3 ] ) are clearly seen in this part of the curve corresponding to the phase transitions at @xmath49 k and @xmath50 k @xcite .
it is interesting to compare the above measurements to the magnetocaloric effect in another frustrated spin system @xmath1 .
the data on @xmath1 have been obtained on a sintered powder sample ( mass 4.05 g ) using an experimental technique similar to that described above . due to the large heat capacity of the sample , no correction for parasitic heat leaks
is required in this case .
the measured adiabatic demagnetization curves for @xmath1 are very similar to the previously published results for oriented single crystals@xcite in the temperature range of overlap .
our results are compared to @xmath0 data in fig .
[ fig3 ] on a field scale normalized to the corresponding values of @xmath10 : 15 koe and 70 koe , respectively .
the magnetocaloric effect in the @xmath1 sample is qualitatively similar to what is observed in @xmath0 , but occurs in a different temperature range due to a smaller exchange constant . in a demagnetization process of @xmath1 starting from @xmath51 ( 28 koe ) at
@xmath52 k the temperature decreases by a factor of 5 reaching its minimum value of 0.1 k in the vicinity of the saturation field after which it remains practically constant .
( raw data ) and @xmath1 on a field scale normalized to @xmath10 .
monte carlo simulations are shown by the solid lines ; the dashed line is an overcorrection to the experimental data made using formula ( [ tcor ] ) . ]
our experimental results demonstrate the decisive role of the frustration in producing large adiabatic temperature changes in the two materials . in typical nonfrustrated antiferromagnets ,
magnetocaloric effect in the vicinity of field induced transitions is very small .
for example , adiabatic @xmath53 does not exceed 0.05 k in csni@xmath54fe@xmath55cl@xmath56i for a similar range of fields and temperatures.@xcite for a quantitative description of the above data we have performed a series of monte carlo ( mc ) simulations on classical antiferromagnetic models with only nearest - neighbor exchange .
a detailed account of the application of the mc technique to model adiabatic processes is given in ref . .
quantum spins @xmath5 are replaced by classical @xmath57-vectors ( @xmath58 ) , an approximation which is valid for high enough temperatures @xmath59 .
the only parameter required to compare the mc simulations to the experimental data is the exchange constant @xmath60 , which is estimated from values of the saturation field or the curie - weiss temperature .
expressions for these quantities for nearest - neighbor antiferromagnets on pyrochlore and garnet lattices are @xmath61 respectively ( see , _ e.g. _ , ref . ) .
corresponding values for both compounds have been taken from previous works @xcite and are summarized in table [ tab1 ] .
for both materials , the exchange constants derived from @xmath10 and @xmath2 agree to within an accuracy of @xmath62 % .
the remaining difference may be attributed to weaker dipolar and other interactions , which are responsible for the eventual ordering in @xmath0 below @xmath63 and above @xmath64 koe in @xmath1 .
@xmath10 and @xmath2 and the estimations of the exchange constants for @xmath0 and @xmath1 . [ cols="^,^,^,^,^",options="header " , ] [ tab1 ] the results of the mc simulations for @xmath0 with @xmath28 k are indicated by the solid lines in figs .
[ fig2 ] and [ fig3 ] . for the scans starting at temperatures above 10
k one can not neglect the phonon contribution to the total heat capacity of the sample .
the influence of lattice heat on the demagnetization process can be taken into account using eq .
( [ balance ] ) , which yields the following relationship between the ideal temperature variation @xmath53 and its observed value @xmath65 : @xmath66 where @xmath67 is the phonon part of the specific heat , @xmath42 is its magnetic part . using the results shown in fig .
[ fig1 ] , one can approximately correct the @xmath31 curves for the temperature dependent phonon contribution ( see dotted lines on fig .
[ fig2 ] ) . with this correction ,
the mc calculations demonstrate very good agreement with the experiment in the whole temperature range at fields above @xmath10 .
since a nearest - neighbor heisenberg pyrochlore antiferromagnet does not order , the mc simulations can not describe satisfactorily the observed behavior below @xmath10 at @xmath68 k. besides , this is also the limit of the validity of the classical approximation : @xmath69 k. the temperature increase at @xmath70 observed in @xmath0 ( fig .
[ fig3 ] ) may be attributed to the reopening of a gap in the excitation spectrum below @xmath10 caused by anisotropic interactions .
the theoretical dependence obtained for @xmath1 ( lower solid line in fig . [ fig3 ] ) with the exchange constant @xmath71 k also fits the high field part of the experimental adiabatic curve very well . under isothermal demagnetization from @xmath72 koe to @xmath73 at different temperatures .
entropy variations of an ideal @xmath5 paramagnet under the same conditions are shown by the dashed lines .
the inset represents the critical intersection line of the two surfaces .
] finally , we estimate the cooling power of the demagnetization process using the measured adiabatic @xmath31 curves along with the specific heat data @xmath16 obtained at high fields . the amount of heat absorbed by a magnetic material during isothermal demagnetization is related to the entropy change @xmath74 .
consider an adiabatic demagnetization curve , which starts at @xmath75 and ends at @xmath76 .
the entropy along the curve @xmath31 remains constant , so we can relate the entropy variations at constant temperature and in constant field :
@xmath77 the entropy changes @xmath78 of @xmath0 under the isothermal demagnetization at various temperatures from the initial field @xmath72 koe as a function of the final field @xmath73 are presented by closed circles in fig . [ cool ] .
the values of @xmath78 are calculated from our experimental data using the above equation , while the real entropy of the system remains undefined .
two features should be underlined : ( i ) about one half of the total magnetic entropy @xmath79 remains in the system even at temperatures very close to the transition into an ordered state at @xmath17 k ; ( ii ) the largest entropy change @xmath78 and consequently , a heat absorption @xmath80 occurs in a high field region above @xmath10 .
this differs significantly from the behavior of an ideal paramagnet at low temperatures ( shown by grey surface on fig .
[ cool ] ) , which releases a significant part of its entropy only if demagnetized to @xmath81 .
this figure demonstrates that @xmath0 has a considerable advantage below 5 k in the _ high field _
cooling power over conventional paramagnets .
the boundary of the field - temperature area where the pyrochlore is advantageous over an ideal @xmath5 paramagnet obtained as an intersection of the two sheets ( the bold line ) is shown in the inset to fig .
[ cool ] . for comparison
, we performed similar mc simulations of a nearest - neighbor antiferromagnet ( not shown in the graph ) .
classical models can not give correct values for the total entropy of a quantum spin system .
nevertheless , the variation of entropy is reproduced with a remarkable accuracy .
for example , the calculated values of @xmath78 at @xmath82 k for demagnetization from @xmath72 koe to zero field and to @xmath83 koe are 0.52 and 0.40 of the total magnetic entropy respectively .
the difference from the experiment does not exceed 20% , which is due to the large @xmath5 spin of the gd@xmath4 ions .
the cooling power @xmath80 has a maximum around 4 k reaching 30 j / mole gd .
such an amount of heat corresponds , for comparison , to the evaporation heat of approximately one mole of liquid @xmath13he at @xmath84 k. the garnet compound @xmath1 is also advantageous as a refrigerant material over paramagnetic salts , but at lower fields and temperatures .
we suggest that a combination of the two compounds might become the basis for a inexpensive two - stage adiabatic demagnetization refrigerator suitable for effective cooling from @xmath85 k down to 100 mk range in a single field sweep .
in conclusion , a large magnetocaloric effect is observed in the frustrated pyrochlore antiferromagnet @xmath0 in agreement with recent theoretical predictions.@xcite this observation points at the presence of a macroscopic number of local low - energy excitations in the vicinity of @xmath10 .
such modes can be directly probed in quasielastic neutron scattering measurements analogously to the experiment on zncr@xmath8o@xmath9.@xcite these excitations may also be responsible for the nonfrozen spin dynamics below @xmath63 observed in @xmath0 by muon spin relaxation measurements .
@xcite a comparison between our experimental data and classical mc simulations shows that this numerical technique can semiquantitatively predict the magnetocaloric properties of real rare - earth materials with large ( semiclassical ) magnetic moments , which have been described so far only in the molecular field approximation .
the authors are grateful to v. i. marchenko and m. r. lees for valuable discussions .
this work is supported by rfbr grant 04 - 02 - 17294 and by the rf president program .
is also grateful to the national science support foundation for the financial help .
30 j. villain , z. phys .
b * 33 * , 31 ( 1979 ) . j. t. chalker , p. c. w. holdsworth , e. f. shender , phys
lett . * 68 * , 855 ( 1992 ) .
r. moessner and j. t. chalker , phys .
80 * , 2929 ( 1998 ) .
b. canals and c. lacroix , phys . rev .
b * 61 * , 1149 ( 2000 ) .
a. p. ramirez , geometrical frustration , in _ handbook of magnetic materials _ vol .
13 , edited by k. h. j. buschow ( elsevier , amsterdam , 2001 ) . n. p. raju , m. dion , m. j. p. gingras , t. e. mason , and j. e. greedan , phys .
b * 59 * , 14489 ( 1999 ) . j. d. m. champion , a. s. wills , t. fennell , s. t. bramwell , j. s. gardner , and m. a. green , phys . rev .
b * 64 * , 140407(r ) ( 2001 ) . a. p. ramirez , b. s. shastry , a. hayashi , j. j. krajewski , d. a. huse , and r. j. cava , phys .
lett . * 89 * , 067202 ( 2002 ) . s .-
lee , c. broholm , w. ratcliff , g. gasparovic , q. huang , t. h. kim , and s .- w .
cheong , nature * 418 * , 856 ( 2002 ) .
m. e. zhitomirsky , phys .
b * 67 * , 104421 ( 2003 ) .
m. e. zhitomirsky and h. tsunetsugu , phys .
b * 70 * , 100403(r ) ( 2004 ) .
r. a. fisher , g. e. brodale , e. w. hornung , and w. f. giauque , j. chem . phys . * 59 * , 4652 ( 1973 ) ; e. w. hornung , r. a. fisher , g. e. brodale , and w. f. giauque , _ ibid_. * 61 * , 282 ( 1974 ) ; g. e. brodale , e. w. hornung , r. a. fisher , and w. f. giauque , _ ibid_. * 62 * , 4041 ( 1975 ) . j. a. barclay and w. a. steyert , cryogenics * 22 * , 73 ( 1982 ) .
b. daudin , r. lagnier , b. salce , j. magn . magn . mater . * 27 * , 315 ( 1982 ) .
t. numazava , k. kamiya , t. okano , and k. matsumoto , physica b * 329 - 333 * , 1656 ( 2003 ) .
g. balakrishnan , o. a. petrenko , m. r. lees , and d. mck .
paul , j. phys . condens .
matter * 10 * , l723 ( 1998 ) .
p. bonville , j. a. hodges , m. ocio , j. p. sanchez , p. vulliet , s. s. sosin and d. braithwaite , j. phys .
: condens .
matter * 15 * , 7777 ( 2003 ) .
o. a. petrenko , m. r. lees , g. balakrishnan , d. mck .
paul , phys .
012402 ( 2004 ) . j. wolf , k. kiefer , m. c. rheinstdter , k. knorr , and m. enderle , eur .
j. b * 22 * , 461 ( 2001 ) .
w. p. wolf , m. ball , m. t. hutchings , a. f. g. wyatt , and m. j. m. leask , j. phys .
b1 * 17 * , 443 ( 1962 ) ; s. hov , h. bratsberg , and a. t. skjeltorp , j. magn .
* 1518 * , 455 ( 1980 ) .
a. yaouanc , p. dalmas de rotier , p. bonville , j. a. hodges , p. c. m. gubbens , c. t. kaiser and s. sakarya physica b * 326 * , 456 ( 2003 ) . | an adiabatic demagnetization process is studied in @xmath0 , a geometrically frustrated antiferromagnet on a pyrochlore lattice .
in contrast to conventional paramagnetic salts , this compound can exhibit a temperature decrease by a factor of ten in the temperature range below the curie - weiss constant .
the most efficient cooling is observed in the field interval between 120 and 60 koe corresponding to a crossover between saturated and spin - liquid phases .
this phenomenon indicates that a considerable part of the magnetic entropy survives in the strongly correlated state . according to the theoretical model , this entropy
is associated with a macroscopic number of local modes remaining gapless till the saturation field .
monte carlo simulations on a classical spin model demonstrate good agreement with the experiment .
the cooling power of the process is experimentally estimated with a view to possible technical applications .
the results for @xmath0 are compared to those for @xmath1 , a well - known material for low temperature magnetic refrigeration . |
a well documented and researched issue in the context of vortex phase diagrams @xcite of both low @xmath5 @xcite as well as high @xmath5 superconductors @xcite is the phenomenon of peak effect ( pe ) @xcite , which is an anomalous increase in the critical current density ( @xmath6 ) prior to reaching the normal - superconductor phase boundary .
the pe is widely considered to mark a first order transition from a collective pinned ordered vortex solid to an individually pinned disordered solid @xcite .
another anomalous feature seen deep in the mixed state of weakly pinned samples of low as well as high @xmath5 superconductors is the second magnetization peak ( smp ) anomaly @xcite .
the smp anomaly is often related to the pinning induced transition from dislocation free elastic glass ( bragg glass @xcite ) to the dislocation mediated multi - domain vortex glass state . in such a circumstance ,
the onset field of the smp anomaly is not expected to vary with temperature @xcite . + abrikosov @xcite had predicted the flux lines to be arranged in a regular array , he found the periodic array to be a square , but the difference in energy between the square and a triangular array is only 2 @xmath7 .
numerous reports of change in the symmetry of the vl from a rhombohedral towards a square symmetry @xcite observed by small angle neutron scattering ( sans ) measurements and bitter decoration studies @xcite in the quaternary borocarbide compounds @xcite , have resurged the interest in exploring and understanding the underlying mechanism governing this process in a wide variety of superconductors , e.g. , * pb*tl @xcite , @xmath8 @xcite , @xmath9 @xcite , nb @xcite , @xmath10 @xcite , etc .
+ the borocarbide superconductors which have a tetragonal crystal structure with c / a @xmath11 3 , are convenient test beds to study the interesting phenomenon of the change in the flux line lattice symmetry and the effects dependent on it , as high quality single crystals of large enough sizes can be grown by different procedures @xcite . in the non - magnetic members of the borocarbide series , viz .
, @xmath0 ( y1221 ) and @xmath4 ( lu1221 ) , it is observed that for h @xmath2 [ 001 ] at low fields ( @xmath11 0.1 t ) , the vl symmetry is a distorted triangle ( lower field rhombohedral @xmath12 ) with an apex angle @xmath13 . with the increase in field , the vl undergoes a sudden ( first order ) transformation via a @xmath14 re - orientation to higher field rhombohedral @xmath15 , with an apex angle @xmath16 ( @xmath17 ) . the @xmath15 symmetry
subsequently smoothly proceeds to a square symmetry via a continuous ( second order ) transition at a field @xmath18 @xcite . in the typical crystals of y1221 , the @xmath19 field ( for h @xmath2 [ 001 ] )
lies in the range @xmath11 100 - 150 mt @xcite , whereas in the very clean crystals of lu1221 , the same field lies in the lower field interval , 20 - 50 mt for h @xmath2 c @xcite .
further studies @xcite in crystals of @xmath4 have revealed that for h @xmath2 [ 100 ] , the rhombohedral @xmath15 ( apex angle @xmath20 ) undergoes a sudden reorientation transition at @xmath21 300 mt , such that the body diagonal of distorted rhombohedral locks up in [ 010 ] direction with apex angle @xmath22 .
+ kogan _ et al .
_ @xcite predicted the occurence of the discontinous ( first order like ) @xmath23 transition at @xmath11 20 mt for h @xmath2 [ 001 ] in clean crystals of borocarbides .
subsequent studies by gammel _
@xcite in co doped lu1221 crystals revealed that the @xmath15 to square transition for h @xmath2 [ 001 ] shifts from lower to higher fields with the increase in the co doping , i.e. , with the increase in disorder effects .
however , the possibilities of relationship(s ) between the structural transition(s ) in the vl and the spatial order - disorder transitions _ a la _ pe / smp have not been described .
it is of interest to know ( i ) how the symmetry transformations adjust to the pinning landscape , and ( ii ) whether the domain volume within which vl remains correlated depends on the underlying symmetry and the disorder effects on it .
magnetization studies by silhanek _
@xcite on a single crystal of y1221 with an applied field h @xmath2 c ( i.e. , [ 001 ] ) revealed the presence of a kink in the pinning force density at a field value , which is close to the @xmath19 value in this compound . to our knowledge , no other signature(s ) of the vl symmetry transitions
have been reported in the magnetization hysteresis measurements , while there have been several reports of the observation of the pe in the samples of y1221 @xcite and lu1221 @xcite .
+ we report here on the observation of the pe , the smp anomaly and the flux jumps in the same isothermal magnetization hysteresis scan for h @xmath2 c in the crystals of y1221 and lu1221 .
the flux jumps interestingly occur in the field regime , where the vl symmetry transition is reported to occur across the respective @xmath19 values for applied field oriented in the c - direction in each of the compounds .
we believe that these flux jumps indeed have a correlation with the local symmetry change in the vortex lattice of borocarbide superconductors . to corroborate this possibility ,
we have traced in several ways , the minor magnetization curves by changing the initial thermomagnetic history of the sample of y1221 .
the measurement of the quadrupolar signal , which purports to preferentially fingerprint the inhomogeneity in the magnetization across the sample , also , registers the change in the symmetry of the vl .
the loci of the threshold fields at which the smp anomaly and the pe commence can lead to the demarcation of boundaries across which changes in the spatial correlation of vl occur .
we present construction of the vortex phase diagrams for h @xmath2 c in the crystals of y1221 and lu1221 .
the parametric region in ( h , t ) space over which elastic glass state exists , seems to be influenced by the purity of the crystal . in y1221 crystal ,
elastic glass state spans over a smaller region , indicating stronger pinning effect in this sample as compared to that in the crystal of lu1221 .
magnetization measurements have been performed on two single crystals of y1221 , labeled as a and b , and a crystal of lu1221 . the single crystals , a and b , of y1221 were grown by the travelling solvent floating zone method @xcite , while the single crystal of lu1221 was grown by the flux method , using @xmath24 as flux @xcite .
the crystal a of y1221 and that of lu1221 are ( thin ) platelets in shape , with the c - axis perpendicular to the plane of the platelet .
the crystal b of y1221 is , however , a parallelopiped ( of size @xmath11 3 mm ( l ) @xmath25 0.7 mm ( b ) @xmath25 0.67 mm ( t ) ) in shape , with the a - axis along the largest dimension .
both the crystals of y1221 have @xmath26 @xmath27 , whereas the crystal of lu1221 has a transition temperature , @xmath5(0 ) @xmath28 .
the dc magnetization measurements were performed using ( i ) a 12 tesla vibrating sample magnetometer ( vsm ) ( oxford instruments , u.k . ) and ( ii ) a 7.5 tesla squid magnetometer ( model mpms7 , quantum design inc . , u.s.a . ) .
a ) with h @xmath2 c. the presence of a peak effect ` bubble ' and flux jumps ( four in the third quadrant and three in the fifth quadrant ) can be noted in the panel ( a ) .
panel ( b ) shows a blow up of the m - h loop highlighting the second magnetization peak anomaly .
the locations of the @xmath29 , @xmath30 and @xmath31 are marked in this panel . ]
the pe in a weakly pinned crystal a of y1221 is clearly evident ( see fig .
1(a ) ) by the characteristic ` bubble ' in a five quadrant m - h loop recorded at 2.1 k using a vsm with a scan rate of 0.35 t / min and with an applied field h @xmath2 c. also , evident in fig . 1 ( a ) is the presence of flux jumps at fields less than 200 mt , far below the peak effect region .
the field values at which the flux jumps occur , vary somewhat from scan to scan , recorded at the same ramp rate of the swept magnetic field .
this phenomenon will be described in detail in section b. + fig .
1(b ) shows the m - h plot on an expanded scale to emphasize the presence of an anomalous feature designated as the second magnetization peak ( smp ) anomaly around a field of 1 t. the large hysteresis in the magnetization at fields less than 0.5 t in fig .
1(a ) gives way to a much smaller irreversibility at higher fields ( upto close to the onset field @xmath29 of the pe ) , indicating that the pinning in the sample is weak and an ordered elastic glass phase ( i.e. , a bragg glass ( bg ) @xcite like state ) gets established well before the field regime of the pe . from the hysteresis plot of fig .
1(a ) , it may seem that an ordered bg phase extends all the way up to onset field of the pe , but a closer look at the data in fig .
1(b ) ( which reveals the presence of a smp anomaly ) , suggests that the bg phase could terminate at the onset field of the smp anomaly above which the multi - domain vortex glass ( vg ) phase ensues @xcite .
the hysteresis width , @xmath32(h ) ( @xmath33(h ) ) , starts to decrease with field once again above the peak field of the smp anomaly and continues upto the onset field of pe , thereby implying an improvement in the spatial order in this interval .
a complete amorphization of the multi - domain elastic vl commences only at @xmath29 .
( h)/@xmath6(@xmath30 ) ( = @xmath34m(h)/@xmath34m(@xmath30 ) ) vs. field ( @xmath35h ) in y1221 ( @xmath36 a ) at two temperatures ( 2.1 k and 6.5 k ) for h @xmath2 c. the locations of the maximum position of the smp anomaly ( @xmath37 ) and the onset ( @xmath29 ) and maximum ( @xmath30 ) positions of the pe are marked . ] fig .
2 shows a ( normalized ) @xmath6 versus h plot corresponding to the data of fig .
also , shown for comparison , is a ( normalized ) @xmath6(h ) plot ( h @xmath2 c ) , at a higher temperature ( t = 6.5 k ) .
the peak fields of the pe ( @xmath30 ) and the smp ( @xmath37 ) stand marked for both the temperatures .
@xmath30 decreases considerably with an increase in temperature ( from 5.4 t at 2.1 k to 3.7 t at 6.5 k ) , but @xmath37 remains near 1.1 t at both the temperatures .
this leads to an inference that the @xmath37(t ) line in y1221 ( @xmath36 a ) would exhibit a very weak temperature dependence , reminiscent of the behavior of the smp anomaly in another weakly pinned low @xmath5 superconductor , viz .
, @xmath38 ( @xmath39 8.2 k ) @xcite .
m/@xmath34@xmath40(@xmath41 ) vs. normalized field ( h = h/@xmath31 ) in y1221 ( @xmath36 a ) and lu1221 at t = 6.5 k for h @xmath2 c. the peak effect as well as the second magnetization peak anomaly are evident in the data of both the crystals . while only one power law region is evident for the case of lu1221 , two power law regions , viz .
, pl - i and pl - ii can be seen for y1221 . ]
3 shows a log - log plot of the current density ( normalized to its peak value across the pe , where the correlation volume of vl is expected to reach a minimum ) with field ( normalized to the upper critical field @xmath31 ) in y1221 ( @xmath36 a ) with h @xmath2 c and at 6.5 k. also , shown for comparison in this figure , is a plot in lu1221 , at ( nearly ) the same temperature with h @xmath2 c. both compounds exhibit smp like anomaly distinct from the pe feature , but with one little difference . while for y1221
, the onset field of smp anomaly lying deeper inside the mixed state is well separated from the pe feature and does not display temperature variation , for lu1221 , the smp anomaly lies at the edge of the pe and varies with temperature as the pe .
+ the field regime where @xmath42 varies with field in a power law manner , i.e. , @xmath43 , is often demarcated @xcite as the collective pinned elastic regime . in lu1221 ,
the power law behavior can be seen ( cf .
3 ) to extend upto h (= h/@xmath31 ) @xmath11 0.6 , after which the bg phase probably gets broken up into multi - domain vg phase , which starts to amorphize at the onset of pe ( h @xmath11 0.85 ) .
on the other hand , for y1221 , one can mark out two regions for the power law behavior .
the first power law region ( pl i ) extends only upto h @xmath11 0.18 and the second region ( pl ii ) surfaces between h @xmath11 0.4 and 0.75 , where the dislocations injected in the interval , h @xmath11 0.18 - 0.4 , could partially heal . + the ratio of @xmath6 at the peak position of the pe to that at the onset of the pe is about five for y1221 , while the same ratio is about two for lu1221 .
this implies that the correlation volume in y1221 shrinks to about 1/25 of its value at the onset of pe ( @xmath6 @xmath44 1/@xmath45 ) , while for lu1221 , @xmath46 at @xmath30 has shrunk only to 1/4 of its value at @xmath29 .
this is plausible , since for lu1221 , the process of reduction in @xmath46 starts at @xmath47 and it continues till the arrival of @xmath29 . on the other hand , for y1221 , the process of reduction in @xmath46 starts at the onset of pe .
prior to it , the vl in this case has a possibility to heal between @xmath37 and @xmath29 .
in fact , the ratio of @xmath6 at @xmath30 to its value at @xmath48 ( h @xmath11 0.6 ) is also about five in lu1221 .
the premise that the smp anomaly is disorder induced could imply that its onset at h @xmath11 0.18 in the given y1221 crystal ( @xmath36 a ) as compared to the onset of smp at h @xmath11 0.6 in lu1221 crystal , reveals the relative purity ( levels of effective disorder ) of the samples of these two compounds .
4 shows a plot of the pinning force density ( @xmath49 h ) with field on a linear - log scale in y1221 ; the data correspond to the magnetization hysteresis measurements shown in fig .
three peaks , marked as @xmath50 , smp and pe , can be clearly distinguished .
an increase in the pinning force upto @xmath50 is a representation of the increase in the rigidity of the vortex lattice as a result of the interactions between the flux lines .
a peak at @xmath50 probably implies that the rigidity of the vortex lattice has attained a limiting value dictated by the interaction effects .
the second hump corresponds to the smp anomaly and the third peak corresponds to the quintessential pe @xcite .
apart from the three maxima , one can also mark out the positions of three flux jumps in this figure .
these jumps are observed to lie close to , though a little lower , the field value corresponding to the @xmath50 .
this suggests that flux jumps are observed in the field regime , where the lattice rigidity is effective and vl is well formed .
@xmath25h ) vs. field in y1221
( @xmath36 a ) with h @xmath2 c. the regimes of @xmath51@xmath52 , smp and pe have been indicated , and the positions of the flux jumps have been marked by arrows . ]
panel ( a ) of fig .
5 shows a five quadrant m - h loop recorded at 2.1 k in a vsm at a sweep rate of 0.25 mt / sec in y1221 ( @xmath36 a ) with h @xmath2 c. one can clearly discern the presence of the multiple flux jumps in selective quadrants .
the following observations are noteworthy : a ) at t = 2.1 k with h @xmath2 c using a vsm with a scan rate of 0.25 mt / sec .
presence of two flux jumps in the third quadrant and three flux jumps in the fifth quadrant can be clearly noted .
panel ( b ) shows the m - h curve obtained after cycling the field several times between @xmath53@xmath54 (= @xmath53 300 mt ) .
note that the number of flux jumps in third and fifth quadrant is now same . ]
\(i ) flux jumps are absent in the zfc run ( 0 @xmath55 ) , in the second quadrant ( @xmath56 ) and in the fourth quadrant ( -@xmath57 ) .
they occur only in third ( 0 @xmath58 ) and fifth ( 0 to @xmath54 , subsequent to the initial zfc run ) quadrant .
+ ( ii ) two flux jumps occur at about -55 mt and -120 mt , in third quadrant and three flux jumps happen at 70 mt , 110 mt and 150 mt in fifth quadrant .
+ on cycling the field repeatedly between @xmath53 300 mt several times , it was noted that the number of flux jumps in third and fifth quadrants stabilized to three ( see panel ( b ) in fig . 5 ) , however , the precise field values at which the jumps happen were found to vary each time . in order to overcome an apprehension that these flux jumps could be an artefact of the rapid ramping of the magnetic field and/or the procedure of magnetization measurement in a vsm , m - h loop was also recorded using a squid magnetometer , where the superconducting magnet is kept in the persistent mode while ascertaining the magnetization value of the sample .
qualitatively , the same behaviour ( data not shown here ) as depicted in fig .
5 was noted .
the number of flux jumps in one of the quadrants ( namely , third ) were , however , observed to increase from three to four .
we are inclined to surmise that the number of jumps in the third / fifth quadrant could statistically vary between two to four .
b ) at t = 2.1 k with h @xmath2 a. ] to establish the notion that the flux jumps do not depend on any specific physical characteristic of a given sample of y1221 , the m - h loops were recorded on another sample of y1221 ( crystal b , which is parallelopiped in shape ) with field applied parallel to its longest physical dimension ( h @xmath2 a in this case , see fig . 6 ) .
the presence of flux jumps in the third and fifth quadrants attests to the fact that demagnetization factor of the sample does not influence the manner in which the flux jumps get observed .
the observation of flux jumps for h @xmath2 a orientation in fig .
6 , also , suggests that the mechanism responsible for these jumps probably does not depend on the orientation of the applied field w.r.t . the crystalline axis
. it may be pertinent to state here that flux jumps in the third and fifth quadrants have been noted in the hysteresis loops pertaining to the intermediate state of some of the specimen of type - i superconductors , viz .
, pb @xcite .
it is believed that in such a circumstance , the flux jumps are caused by the escape of flux by the fusion of local macroscopic regions of positive and negative magnetization lying in juxtaposition .
+ in order to comprehend the mechanism governing the flux jump process , field profiles relevant to fig .
5 are drawn , as per a simplified prescription of the bean s critical state model @xcite for an infinite slab of thickness 2d , with surfaces at x = @xmath53 d , the applied field h being parallel to the surface .
we realize that our experimental sample is either platelet shaped or is a parallelopiped , but it is hoped that the bean s profile drawn in fig . 7 , for zero demagnetization limit and for @xmath6 ( h ) = constant for the sake of simplicity , would qualitatively turn out to be instructive even when @xmath6 ( h ) decreases with an increase in h. for brevity ,
profiles are shown in only one half of the sample .
the other half is a mirror image of the drawn profiles .
while drawing bean s field profiles , b - x ( local macroscopic field , b , versus distance , x , from the centre of the sample ) , we have assumed that the full penetration field , @xmath59 , is 200 mt . in the prescription of bean s model , @xmath59 is the limiting field at which the applied field invades the entire sample , after initial zero field cooling the sample ( see panel ( a ) of fig .
the estimate of @xmath11 200 mt for @xmath59 in y1221 , with h @xmath2 c , at 2.1 k is based on the magnetization data in fig . 5 ( a ) .
it is the limiting field where the virgin zfc magnetization curve would merge with the forward leg of the envelope @xcite loop in the absence of the flux jumps .
when @xmath6 ( h ) decreases with field , such a limiting field can be taken to identify the nominal @xmath59 value .
+ in fig . 7 , we have restricted @xmath53 @xmath54 to @xmath53 300 mt in conformity with the m - h data shown in fig .
panels ( c ) to ( e ) of fig .
7 are relevant for the m - h loop shown in panel ( b ) of fig . 5 as well .
it may be noted that the field profiles in panels ( a ) , ( b ) and ( d ) of fig .
7 are such that the magnetic field inside the sample remains of the same sign ( positive or negative ) in these cases .
the field profiles in panels ( c ) and ( e ) of fig . 7 allow for the possibility of both positive and negative field values inside the sample .
the positive and negative ( macroscopic ) fields may be identified with the domains ( or local regions ) comprising vortices and anti - vortices .
a narrow region encompassing zero field value would be free of any kind of vortices , and in its neighborhood , the domains of vortices and anti - vortices would lie in juxtaposition .
the fact that the flux jumps are observed in the third and the fifth quadrants in conjunction with the bean s profiles in fig .
7 could further imply that co - existence of domains of vortices and anti - vortices is necessary for the jumps to occur .
+ we would like to now surmise that mere juxtaposition of the regions of vortices and anti - vortices is not adequate to trigger flux jumps .
we conjecture that sudden annihilation of vortices ( flux jumps ) gets triggered , when a change in the local symmetry of the vortex lattice from low field rhombohedral @xmath12 to higher field rhombohedral @xmath15 ( or vice - versa ) occurs over a portion of the sample ( i.e. , in some domains ) , while the regions of positive and negative fields lie in juxtaposition anywhere else in the sample .
let us call the domains comprising rhombohedral @xmath12 and negative(positive ) field as @xmath60 , and , similarly , the domains with rhombohedral @xmath15 and negative(positive ) field as @xmath61 .
the @xmath62 ( and vice - versa ) transition(s ) can be expected to happen in y1221 crystal as the ( macroscopic ) field inside the sample crosses the region of 100 - 150 mt during the ramping of the external field .
in addition , the @xmath15 domains can , however , exist in ( metastable ) supercooled state during the field cooling process even in a field value somewhat less than 100 mt , as has been reported in some of the sans experiments @xcite .
such metastable domains could display the tendency to suddenly transform to the stable domains , during the subsequent field ramping cycle .
+ on the basis of the above conjecture , let us now reexamine the observations in fig . 5 . during the first ramp up of the field after zfc ( 0 mt to 300 mt ) , the disordered bundles of vortices will enter the sample from the corners and edges and attempt to settle down into regions of @xmath63 and @xmath64 domains , with the former lying near the centre for h = 300 mt ( see fig .
7 ( a ) ) . as the applied field
is reduced to 0 mt ( see fig . 7 ( b ) ) , the @xmath63 domains would be present near the sample edge and @xmath64 domains are likely to lie in the interior .
no flux jumps happen in the first two quadrants , as the domains with anti - vortices are not present . in the third quadrant , as the field gets cycled to negative values , the first flux jump happens near about -50 mt . the field profiles in fig . 7 ( c )
imply that in such a circumstance , the @xmath65 domains would lie near the edge , while @xmath63 and ( supercooled / metastable ) @xmath64 domains could exist in the interior of the sample .
a change in the symmetry from @xmath66 in some domains could trigger an additional perturbation in the dynamically varying distribution of @xmath65 , @xmath63 and @xmath64 domains , such that the vortices and anti - vortices lying in juxtaposition in @xmath63/@xmath65 domains could start annihilating each other leading to an avalanche resulting in a flux jump .
= -100 mt corresponding to a situation ( a ) just before the flux jump and ( b ) just after the flux jump .
( c ) bean s profiles as the field ramps upto @xmath35h = -140 mt ] from fig . 5
, one can find that the magnetization values start to build up once again after the flux jump happens , and one can envisage the notion of an underlying envelope m - h loop . as the applied field ramps up beyond -100 mt , another ( more ) flux jump(s ) happen ( cf . fig . 5 ( b ) ) .
such jumps could possibly be triggered by @xmath67 transition in the domains , which are nucleating near the edge of the sample .
when the applied field is in the range of -100 mt to -150 mt , the @xmath63 domains would be present in the middle of the sample .
their presence would permit the annihilation of vortices and anti - vortices in @xmath63 and @xmath65 domains lying in juxtaposition in the interior .
the recurrence of the process of @xmath67 transition could trigger more than one avalanche , while the field is getting ramped down to -200 mt in the third quadrant .
similar arguments will explain the flux jumps in the fifth quadrant .
+ we show in fig . 8
a plausible sequence of bean s profile , when the flux jump is located , say at h = 100 mt . as the flux jumps occurs , the bean s profile of fig .
8 ( a ) probably transforms to the profile in fig .
8 ( b ) , due to movement of anti - vortices from the exterior of the sample into the portion in which anti - vortices and vortices have annihilated to create a current free region at the right side of b = 0 .
the profile in fig .
8 ( b ) implies the rearrangement of vortices in the interior of the sample , as shown by the solid line .
it is apparent that profile in fig .
8 ( b ) corresponds to a lower net magnetization value as compared to that for profile in fig . 8 ( a ) . as the field would further ramp away from -100 mt
, the critical current would gradually get set up in the gradient free region .
fig 8 ( c ) shows that at -140 mt , the bean s profile has assumed a form , as if the flux jump had not occurred at -100 mt , thereby implying the return of the magnetization values to the underlying envelope hysteresis loop . to check the validity of our conjecture
, we traced several minor hysteresis curves with different thermomagnetic histories , including the tracings of the complete hysteresis loops after having cooled the sample in different fields .
9 shows two representative minor hysteresis curves obtained at 2.1 k in y1221 ( @xmath36 a ) using the field ramp rate of 0.2 t / min in the vsm . in panel
( a ) of fig . 9 , after zero field cooling
, the field is initially ramped upto + 75 mt ( filled squares ) , it is then reversed to -500 mt ( open triangles ) , followed by ramping up again to + 75 mt ( filled circles ) .
no flux jump is observed at any magnetic field in the third quadrant , however , a flux jump occurs at about + 32 mt in the fifth quadrant .
+ panel ( b ) of fig .
9 shows that if the field is reversed to -500 mt from + 174 mt , two flux jumps are observed at about -91 mt and -137 mt , respectively , in the third quadrant . it is pertinent to note that a flux jump in the range of -30 mt to -50 mt is not present in the third quadrant . on ramping up the field from -500 mt to + 174 mt
, one can witness three flux jumps at about 37 mt , 97 mt and 142 mt , respectively , in the fifth quadrant .
a ) with h @xmath2 c. the open triangle and filled circle data points in different panels correspond to the minor curves traced while ramping up the field to -500 mt from ( a ) 75 mt and ( b ) 174 mt respectively , and reversing the field from -500 mt to ( a ) 75 mt and ( b ) 174 mt , respectively . ]
10 shows the bean s profiles corresponding to the data in fig .
the profiles in fig .
10 ( a ) imply that when the applied field reaches a value of -75 mt , the @xmath65 and @xmath63 domains lie in juxtaposition in the outer portion of the sample and there are no vortices of any kind in the middle region of the sample .
as the applied field gets ramped from -75 mt to -150 mt , the specimen gets progressively more filled with @xmath65 domains .
as the ramping of the field proceeds further from -150 mt to -500 mt , the @xmath65 to @xmath68 transition could ( repeatedly ) occur near the edge of the sample , but there are no domains with vortices of the opposite kind ( @xmath63 like or @xmath64 like ) in the interior .
the conjecture stated above thus , precludes the occurrence of flux jump in the third quadrant in fig .
the profiles in fig . 10 ( b ) further show that when the flux jump happens on approaching + 37 mt field in fig . 9 ( a ) , the @xmath63 and @xmath65 domains would lie in juxtaposition near the edge of the sample and @xmath68 domains would lie in the deep interior , near the central region .
a transition from @xmath68 to @xmath65 in the interior region probably provides the necessary trigger for annihilation of vortices across the boundaries of @xmath65 and @xmath63 domains .
+ an examination of the profiles in fig .
10 ( c ) in conjunction with the flux jumps in fig .
9 ( b ) would imply that as the applied field reaches a value of -50 mt in the third quadrant , the @xmath65 domain will lie near the edge , and only the @xmath63 domains would fill the interior of the sample .
there is no possibility of @xmath23 transition anywhere in the sample between 0 and -50 mt . the @xmath67 transition probably happens near the edge of the sample as the applied field ramps towards -91 mt ( see fig . 9 ( b ) )
, at that stage , the interior of the sample contains @xmath63 domains . as the field ramps further to -137 mt
, another @xmath65 to @xmath68 transition could occur near the edge , while the @xmath63 domains are still left in the interior of the sample .
this could explain the two flux jumps in the third quadrant in fig .
9 ( b ) . the profiles in fig
. 10 ( d ) further show that as the applied field approaches + 37 mt in the fifth quadrant , the situation is similar to that in the fifth quadrant as depicted in fig .
10 ( b ) and the first flux jump in fig . 9 ( b ) gets triggered at such a field , by @xmath68 to @xmath65 transition in the interior .
the subsequent two jumps in the fifth quadrant in fig .
9 ( b ) are triggered by the possibility of recurring @xmath63 to @xmath64 transitions near the sample edge , while @xmath65 domains continue to exist in the interior . + field cooled ( fc ) measurements , in principle , result in a near uniform field distribution across the sample . in a fc(h ) state , the crystal will comprise @xmath12/@xmath15 or a co - existence of @xmath12 and @xmath15 domains , depending on the field value .
tracings of minor hysteresis loops by changing the field in the third quadrant , from different @xmath69 values revealed ( data not being shown here ) one / two / three jumps .
their occurence can also be rationalized in terms of above stated conjecture by drawing appropriate bean s field profiles .
it is well documented that the tendency and the magnitude of the flux jumps decreases with an increase in temperature @xcite .
the plots of the m - h loops in y1221 ( @xmath36 a ) at higher temperatures ( t @xmath1 2.1 k ) in fig .
11 seem to conform to this notion .
a comparison of the data in fig . 11 with the corresponding data in fig .
5 ( a ) reaffirms the premise that the extent of jump(s ) decreases on progressive increase of temperature from 2.1 k to 8.45 k. a significant observation is that the flux jump evident at lower fields ( 50 - 75 mt range ) in fig .
5 ( a ) is not present at higher temperatures ( i.e. , t @xmath70 4 k as in fig .
this trend could also find a rationalization in terms of the decrease of @xmath6(h ) ( or full penetration field ) with the increase in temperature , and the corresponding changes in the field distribution inside the sample . in fig .
11 ( b ) , the @xmath59 value at 6.45 k would be reckoned to be about 150 mt .
bean s profile drawn for @xmath59 = 150 mt would then demonstrate the absence of flux jumps at the lower field values .
a ) with h @xmath2 c. inset in panel ( c ) shows temperature variation in @xmath71 and @xmath31 ( see text for details ) . ] our assertion that a flux jump is triggered by @xmath23 transition and is facilitated by the juxtaposition of vortices and anti - vortices implies that the ( highest ) limiting field value at which a jump is observed ( in a given fifth / third quadrant ) has to be lower than the corresponding @xmath59 at a given temperature .
such a limit could also be taken as indicative of the field value at which @xmath23 transition happens near the edge of the sample . keeping this in view , we draw attention to the plot of such limiting values ( designated as @xmath71 and marked by arrows in panels ( a ) to ( c ) of fig .
11 ) as a function of temperature in the inset of fig . 11 ( c ) .
multiple values of @xmath71 at a given temperature represent the spread in these values during different m - h runs at the chosen temperatures . it is satisfying to see the similarity in @xmath71(t ) line determined as above from our flux jump data and the @xmath19(t ) line ( representing @xmath23 transition ) determined from sans measurements in a crystal of y1221 by dewhurst _
et al _ @xcite .
c. panels ( a ) and ( b ) depict the m - h curves in the fifth ( ramping the field from 0 to @xmath54 = 200 mt ) and the third ( ramping the field from 0 to -@xmath54 = -200 mt ) quadrants , respectively . the locations of the flux jumps have been marked with arrows . ] the plots in fig .
13 show parts of the m - h loop recorded using vsm at 2.2 k in lu1221 , for h @xmath2 c , with a scan rate of 0.25 mt / sec .
the presence of the three flux jumps in the third and fifth quadrants in the field interval 2 to 25 mt can be noted . from bitter decoration experiments performed in the field
cooled state on single crystals of lu1221 , the @xmath23 transition is reckoned to occur in the interval of 20 mt to 50 mt @xcite .
the observation of flux jumps in a single crystal of lu1221 at smaller field values as compared to those in y1221 is encouraging .
the nominal @xmath59 at 2.1 k in the given crystal of lu1221 , with h @xmath2 c , is about 30 mt .
simplistic reasoning based on bean s profiles for @xmath59 = 30 mt can rationalize the occurrence of the flux jumps at the observed field values .
the local macroscopic field would envisage @xmath64 to @xmath63 transition deep inside the sample and @xmath65 to @xmath68 transition near the surface as the applied field ramps from zero to -30 mt in the third quadrant .
when the magnetization in a given sample is non - uniform , multipole moments other than the dipole , also , contribute @xcite to a measured signal .
preferential measurement of the quadrupole moment ( q ) is a very useful technique to gain information on the spatial inhomogeneity in the magnetization across a given sample @xcite .
such a measurement has been performed using a vsm in y1221 ( @xmath36 a ) , with h @xmath2 c ( see fig .
14 ) to explore the fingerprints of the notion of the flux jumps observed in the m - h loops .
the details of the measurement procedure can be found elsewhere @xcite . in brief , a sample is moved from the central region of the astatic pair of the coils of vsm to another location , where the signal due to the dipole moment is expected to cross from a positive to a negative value . at such a location , the measured signal preferentially captures the contribution from the quadrupolar moment of the sample .
a ) for h @xmath2 c at t = 2 k and 2.9 k , respectively . ] panels ( a ) and ( b ) of fig .
14 show q ( in arbitrary units ) vs. h plots at 2.0 k and 2.9 k , respectively as the field is sweeped from zero to @xmath54 in the so called fifth quadrant at a sweep rate of 0.01 t / minute .
the q vs. h data in fig .
14 ( a ) appears to have a shape similar to the m vs. h data in fig . 5 , and this could imply that the residual contribution from the dipole moment of the sample is still dominating the measured signal at the new preferred location for the record of the quadrupolar contribution .
there is , however , one notable observation in both the panels of fig .
14 , which purports to support the significant presence of the contribution from the quadrupolar moment of the sample at the new location .
it may be mentioned that at the field values of the flux jumps , the measured signal gets out of the range of the plot with the lock - in amplifier settings getting overboard momentarily and it returns to within the range only when the process of the flux jump is complete . a motivated search for such an occurrence ( i.e.
, signal getting out of range ) in the magnetization response at the usual central position of the astatic pair of the coils , did not yield an affirmative answer .
this implies that the residual presence of the dipolar signal at the preferred position for the q measurement , is not responsible for the signal getting out of range , when the flux jump happens .
+ we are tempted to conjecture that a readjustment in the vortex matter during the flux jump process causes the contribution from the quadrupolar moment to undergo a peak like behavior .
we believe that a study of anomalous variations in the quadrupolar signal , conveniently measurable in a vsm , has the potential to reveal the changes in the state of the vortex matter , which may not get fingerprinted as anomalous variations in the magnetization hysteresis response , like , the smp anomaly and/or the peak effect phenomenon . collating all the data together
, we draw the phase diagram for y1221 ( @xmath36 a ) and lu1221 for h @xmath2 c. the diagram comprises @xmath71 , @xmath48 , @xmath29 and @xmath31 lines . the @xmath48(t ) lines have been drawn over a limited temperature interval over which data pertaining to it are presently available in these crystals .
it can be noted that @xmath48 line for y1221 is flat , while it decreases with temperature for lu1221 .
the former behaviour is reminiscent of the smp line in @xmath38 @xcite and in optimally doped @xmath72 crystals @xcite , while the latter is similar to the smp line in 2h-@xmath73 crystal @xcite .
( @xmath36 a ) and @xmath4with h @xmath2 c. @xmath71 , @xmath48 , @xmath29 and @xmath31 lines have been sketched and different phases have been named .. ] the smp anomaly is believed to mark a transition from a dislocation free elastic bg phase to a multi - domain vg phase @xcite .
the region below @xmath48 and above @xmath71 marks the bg phase . from fig 15
, it is also evident that the bg state spans over a larger ( h , t ) space in lu1221 as compared to that in y1221 , attesting to the fact that lesser residual disorder prevails in the former sample .
to summarize , we have expounded the details of magnetization hysteresis loops in single crystals of @xmath0 for h @xmath2 c. in the high field regime , a second magnetization peak anomaly and the peak effect are observed , while at lower fields ( h @xmath3 200 mt ) , the flux jumps are evident .
the flux jumps have been noted in minor hysteresis curves recorded with differing thermomagnetic histories , in order to explore a probable reason for their occurrence . on the basis of bean s profiles sketched for different thermomagnetic histories
, we have conjectured that the flux jumps get triggered as a result of the structural transitions , @xmath74 , possible in the vortex lattice in the field range 100 - 150 mt for h @xmath2 c in @xmath0 .
similar flux jumps have also been observed in a cleaner crystal ( as compared to those of @xmath0 ) of @xmath4 at lower fields ( h @xmath3 25 mt ) , giving support to an assertion made by us .
another evidence in support of our conjecture could come from the measurement of quadrupole signal ( q ) in @xmath0 using a vsm .
q undergoes a peak like feature at applied field values , corresponding to which the macroscopic field somewhere inside the sample is such that @xmath74 transition happens .
the onset of a smp anomaly at a lower field in the crystal of @xmath0 as compared to that in @xmath4 could be termed as consistent with the occurrence of @xmath23 transition at a higher field in it .
based on all the measurements , vortex phase diagrams for y1221 ( @xmath36 a ) and lu1221 for h @xmath2 c have been drawn , which depict various phases of vortex matter .
our results call for a study to map bean s profiles by local micro - hall bar arrays @xcite in all the five quadrants for different thermomagnetic histories , in samples which display flux jumps .
three of us ( dj - n , adt and dp ) would like to thank the tifr endowment fund for the kanwal rekhi career development support . we would like to thank morten eskildsen ( university of notre dame , u.s.a . ) for many useful inputs and valuable discussions , in particular , his clarifying account of the changes in field profiles , while a flux jump happens .
the @xmath4 crystals were obtained from paul canfield at ames laboratory , iowa state university , ames , u.s.a .
we also gratefully acknwledge rinke wijngaarden ( vrije universiteit , amsterdam , the netherlands ) for kindly sharing their observations on flux jumps in the samples of pb .
t. giamarchi and s. bhattacharya , in high magnetic fields : applications to condensed matter physics and spectroscopy , c. bertier , l.p .
levy and g. martinez , eds . , p314 ( springer verlag , 2002 ) and the references therein .
s. s. banerjee , a. k. grover , m. j. higgins , gautam i. menon , p. k. mishra , d. pal , s. ramakrishnan , t. v. chandrasekhar rao , g. ravikumar , v. c. sahni , s. sarkar , and c.v .
tomy , physica c * 355 * , 39 ( 2001 ) and the references therein .
u. divakar , a. j. drew , s. l. lee , r. gilardi , j. mesot , f. y. ogrin , d. charalambous , e. m. forgan , g. i. menon , n. momono , m. oda , c. d. dewhurst , and c. baines , phys .
* 92 * , 237004 ( 2004 ) .
r. j. cava , h. takagi , h. w. zandbergen , j. j. krajewski , w. f. peck , t. siegrist , b. batlogg , r. b. van dover , r. j. felder , k. mizuhashi , j. o. lee , h. eisaki , and s. uchida , nature ( london ) * 367 * , 252 ( 1994 ) . v. g. kogan , m. bullock , b. harmon , p. miranovic , lj .
dobrosavljevic - grujic , p. l. gammel and d. j. bishop , phys .
b * 55 * , r8693 ( 1997 ) ; v. g. kogan , p. miranovic , lj .
dobrosavljevic - grujic , w. e. pickett , and d. k. christen , phys .
lett . * 79 * , 741 ( 1997 ) .
k. hirata , h. takeya , t. mochiku and k. kadowaki in : _ proceedings of the @xmath75 international symposium on superconductivity _ , edited by h. hayakawa and y. enomoto ( springer - verlag , tokyo , 1996 ) , p. 619 .
p. chaddah , k. v. bhagwat , and g. ravikumar , physica c * 159 * , 570 ( 1989 ) ; p. chaddah , in _ studies of high temperature superconductors _ , edited by a. v. narlikar , vol .
* 14 * , ( nova science publishers , new york , u.s.a . , 1995 ) , | we present magnetization measurements in single crystals of the tetragonal @xmath0 compound , which exhibit the phenomenon of peak effect as well as the second magnetization peak anomaly for h @xmath1 0.5 t
( h @xmath2 c ) . at the lower field ( 50 mt @xmath3 h @xmath3 200 mt ) ,
we have observed the presence of flux jumps , which seem to relate to a structural change in the local symmetry of the flux line lattice ( a first order re - orientation transition across a local field in some parts of the sample , in the range of 100 mt to 150 mt ) .
these flux jumps are also observed in a single crystal of @xmath4 for h @xmath2 c in the field region from 2 mt to 25 mt , which are compatible with the occurrence of a re - orientation transition at a lower field in a cleaner crystal of this compound , as compared to those of @xmath0 .
vortex phase diagrams drawn for h @xmath2 c in @xmath4 and @xmath0 show that the ordered elastic glass phase spans a larger part of ( h , t ) space in the former as compared to latter , thereby , reaffirming the difference in the relative purity of the two samples . |
measuring the small - angle scattering of x - rays from particles embedded in a uniform medium to calculate their size and shape has long been a standard technique in microbiology ( glatter & kratky , 1982 ) .
the scattering of x - rays from dust grains in the interstellar medium ( ism ) was first considered by overbeck in 1965 .
however , the weak x - ray halo generated by small - angle x - ray scattering from dust has only recently become observable ( mauche & gorenstein , 1989 ; mathis _ et al .
_ , 1995 ; predehl & klose , 1996 ) . in the laboratory ,
small - angle x - ray scattering observations can measure the size and shape of the scattering particles quite accurately , but astrophysical applications are still sharply limited by the availability of sufficiently bright x - ray sources and the spatial and energy resolution of x - ray telescopes . however , recent observations using the _ rosat _ position sensitive proportional counter ( pspc ) ( predehl & schmitt , 1995 ) have found x - ray halos around 25 point sources and 4 supernova remnants , using the moderate @xmath0 resolution of this instrument
. the advanced x - ray astrophysical facility ( axaf ) telescope will have @xmath1resolution ( canizares , 1990 ) , and should dramatically increase the number of sources with visible halos . in general the halo formed around an x - ray point source by small - angle scattering depends , in addition to the spatial distribution of the dust , upon many details of the interstellar dust grain population .
the intensity varies linearly with the number of grains along the line of sight , and is highly dependent upon the size distribution of the larger grains , which more efficiently scatter x - rays .
an individual spherical dust grain scatters x - rays with a angular distribution determined by the grain size distribution , composition , and density . given the dust parameters ,
the spatial distribution of the grains can be estimated by comparing this scattered distribution with the observed halo . although it is difficult to extract each dust grain parameter separately , measurements of the halo s intensity as a function of energy and angle from the source can provide useful limits for the allowable dust grain model parameters .
the general theory of x - ray scattering by dust grains has been discussed in detail by mauche & gorenstein ( 1986 ) ( mg86 ) and mathis & lee ( 1991 ) ( ml91 ) .
their results were based on the rayleigh - gans ( rg ) approximation to the differential scattering cross section .
predehl & klose ( 1996)(pk96 ) discussed the possibilities for extracting dust grain parameters by observing x - ray halos with current x - ray telescopes .
they also considered the limitations of the rayleigh - gans approximation and the effects of absorption in the dust grain on the scattered x - rays .
recent observations of x - ray halos ( mathis _ et al . _ , 1995 ;
woo _ et al .
_ , 1994 ; and predehl , 1996 ) , using both the _ rosat _ and _ asca _ telescopes , have found that the observed halo intensity is less than that predicted by the general theory using the rg approximation .
these authors concluded that their results were evidence for `` fluffy '' dust grains which contain many holes and voids .
these results are of particular interest because of the carbon `` crisis '' in the interstellar medium ( ism ) ( cardelli _ et al .
_ , 1996 ) .
observations have suggested that the amount of carbon available to be in dust grains is substantially less than the amount necessary to make the grains which account for the observed extinction .
mathis ( 1996 ; m96 ) has suggested that fluffy dust grains would account for the extinction measurements while reducing the amount of carbon necessary to be locked up in the dust . however , dwek ( 1997 ) pointed out that the composite fluffy dust model ( cfd ) suggested by mathis fails to resolve the interstellar carbon crisis for several reasons : ( 1 ) the visual albedo of the cfd particles is lower than the observed one .
the increased absorptivity of the composite dust grains produces therefore an excess of infrared emission over that observed with the diffuse infrared background experiment ( dirbe ) and far infrared absolute spectrophotometer ( firas ) instrument on board the cosmic background explorer ( cobe ) satellite ( dwek _ et al .
_ , 1997 ) ; ( 2 ) the cfd model does not include polycyclic aromatic hydrocarbons ( pahs ) as a grain constituent . as a result
it does not take the abundance of carbon locked up in pahs or the effect of pahs on the uv and optical extinction into account . in light of these difficulties we re -
examine the theory of x - ray scattering by interstellar dust particles and the evidence for the existence of composite fluffy dust particles in the ism .
we present in 2 a brief review of the theory , describing the rayleigh - gans ( rg ) approximation to the x - ray scattering from a spherical particle and examining the range of its validity . in
3 we present the exact mie solution to light scattering from a spherical particle , and describe the differences between the mie theory and the rayleigh - gans approximation for various astrophysical situations , including the importance of absorption of x - rays in the dust . in
4 we apply the mie theory to the observations of nova cygni 1992 from mathis _ et al . _
( 1995 ) , and discuss the implications for dust models .
the physical situation we consider is the small - angle scattering of x - rays emitted by a bright x - ray source from dust grains in the ism near the line of sight ( los ) , which will create an x - ray halo around the point source .
the scattered angles are small enough , that for astronomical purposes , we can assume that the path traveled by the unscattered primary x - rays is nearly identical to the paths of scattered photons , in terms of their distances and environments .
we discuss this assumption and its limitations in appendix a. the x - ray source s ( see figure [ fig : scat_diag ] ) at distance d from the observer o has an observed flux @xmath2 and spectrum @xmath3 in the @xmath4 energy range , where @xmath3 is normalized to unity over that interval . by a dust particle located at p towards an observer o.
the parameter @xmath5 is a dimensionless quantity denoting the projected distance to the dust particle from o along the line of sight to s.[fig : scat_diag ] ] the emitted x - rays , traveling though the ism scatter from a population of dust grains with number density @xmath6 per h atom in the interval \{@xmath7 } for grain radii in the range @xmath8 .
we assume that the grain size distribution is uniform but that its total number density can vary along the los .
we characterize the spatial distribution of the grains by the function @xmath9 where @xmath10 is the normalized distance from the observer for an actual distance @xmath11 from the observer ; @xmath9 is normalized to unity over the range \{0,1 } finally , the scattered power in a given direction is determined by the differential scattering cross section ( @xmath12 ) , which is a function of x - ray energy , dust grain size , and scattering angle . in the most general form
, we can write the equation for the scattered intensity at angle @xmath13 , in units of photons per @xmath14 per second per steradian as ( see appendix a ) : @xmath15 equation ( [ scat_eq ] ) apparently diverges for scattering near the source , where @xmath16 .
however , near the source the contribution to the flux scattered to the observer comes from relatively large scattering angles , for which ( @xmath17 ) is vanishingly small .
so in practice using an upper limit @xmath18instead of 1 to the position integral for some small @xmath19 is acceptable , since the integral from @xmath18 to 1 is negligible . numerical experiments , with smoothly distributed dust ( @xmath20 ) , a single grain size , and a single x - ray energy , show that the integral ( using the mie solution for the scattering term ) converges to within 1 part in @xmath21 when @xmath22 ( measured in radians ) .
this conclusion holds for dust sizes in the range @xmath23 and x - ray energies in the range @xmath24kev . using this result
, we can modify equation ( [ scat_eq ] ) by changing the upper limit on @xmath25 to be @xmath26 so that : @xmath27 which removes the numerical difficulties . in addition , in this paper we consider only singly - scattered x - rays .
although ml91 developed the theory of multiply - scattered x - ray halos , the brightest x - ray halos occur when the optical depth to scattering is low , @xmath28 ( pk96 ) .
pk96 also found that when the optical depth for scattering was high enough that multiple scattering was important , absorption along the los greatly reduced both the source and halo intensity .
the rayleigh - gans approximation for the differential scattering cross section @xmath29 requires the following two assumptions .
first , that the reflection from the surface of the dust particle is negligible , _
i.e. _ that @xmath30 , where @xmath31 is the complex index of refraction of the dust .
this ensures that the x - ray enters the dust particle instead of being reflected .
the second , more stringent requirement is that @xmath32 where @xmath33 is the wave number of the x - ray and @xmath34 the radius of the grain .
this ensures that the phase of the incident wave is not shifted inside the medium .
we will show how , in this limit and for sufficiently small scattering angles , it is possible for the waves scattered throughout the dust grain to add coherently . in this situation ,
the intensity of the scattered waves is proportional to the number of scattering sites squared ; _
i.e. _ : @xmath35 where @xmath36 is the mass density of the dust grain , and @xmath34 its radius
. a useful expression for the rg approximation assumptions can be derived using the `` drude approximation '' to @xmath37 for a collection of free electrons responding to an oscillating external field .
the drude approximation ( bohren & huffman , 1983 , p. 253 ) is @xmath38 where @xmath39 is the circular frequency of the wave , @xmath40 is the plasma frequency of the medium ( for electron number density @xmath41 , and where @xmath42 is the classical radius of the electron ) .
the @xmath43 term is a damping term due to electronic interactions with impurities or lattice vibrations .
assuming @xmath44 , we can rewrite ( [ drude ] ) to get @xmath45 where @xmath41 is the density of electrons , @xmath42 is the classical radius of the electron , and @xmath46 the wavelength of the photon . combining ( [ drudeapprox ] ) with ( [ rglimit ] ) , and ignoring damping ( _ i.e. _ , @xmath47 ) we get : @xmath48 equation ( [ basiclimit ] ) can be expressed in terms of the dust mass density @xmath36 by assuming that the atomic constituents of the dust have one electron per 2 nucleons , so @xmath49 .
the coherency requirement for the validity of the rg approximation then becomes : @xmath50 a `` rule of thumb '' for normal interstellar dust grains is that the rg approximation is valid only if the energy of the x - ray in kev is significantly larger than the grain radius in @xmath51 m . within this limit
, the rg approximation completely describes the scattering of photons in a simple analytic form .
the exact rg approximation for the differential scattering cross section is given in mg86 , and shown below .
the exact result is in terms of the spherical bessel function @xmath52 , but can be well - approximated by a gaussian function : @xmath53 where @xmath54 and @xmath55 is a function of the energy and size of the dust particle . we see here the @xmath56 dependence described earlier ; the @xmath57 dependence is implicit in the @xmath58 term ( see eq .
[ drudeapprox ] ) .
insight into the mechanism and the maximum angle for coherent scattering can be derived from the fundamental physics of the scattering .
photons scatter from charged particles , and the differential scattering cross section for photons interacting with a collection of charged free particles is given by jackson ( 1975 ) ( eq .
14.111 ) : @xmath59 where the sum is taken over all the charges in the medium . here
@xmath60 is the charge of the @xmath61th particle , @xmath62 is its mass , and @xmath63 is its position relative to the center of the dust particle ( see figure [ dustscatdiag ] ) .
we define @xmath64 , where @xmath65 and @xmath66 are unit vectors in the initial and scattered directions respectively , @xmath67 and @xmath39 are the circular frequencies of the x - ray outside and inside the medium respectively , and @xmath68 is the speed of light .
the initial and scattered polarization is given by @xmath69 and @xmath70 respectively ; however , we will not consider x - ray polarization in this paper and will therefore ignore these terms .
we see immediately that all the scattering is by electrons ; scattering by protons is strongly suppressed because of their large mass .
figure [ dustscatdiag ] diagrams the situation for one particular scattering site in a spherical dust grain .
impinging upon a dust particle of radius @xmath34 , interacting at a point p located at a distance @xmath71 from the center of the particle o , and scattering in the direction @xmath72.[dustscatdiag ] ] equation ( [ jacksoneq ] ) is strictly valid only for free electrons .
however , x - rays are substantially more energetic than the electronic energy bands in the solids so the electrons in the dust grain can be treated as free ( jackson , 1975 , p. 683 ) .
absorption , however , is not entirely negligible even above 1 kev as there are a number of elements with k - shell electrons that can absorb at these energies ( see 3.2 ) .
we can now show how to derive the limiting angle for coherent scattering . if @xmath73 in equation ( [ jacksoneq ] ) , the exponent term reduces to unity .
then the sum is simply @xmath74 , where @xmath75 is the total number of electrons in the dust grain .
physically , this result is due to each electron scattering independently and the resulting waves adding coherently .
the intensity therefore depends upon the _ number _ of electrons in the scattering medium , and not on the _ spacing _ or _ density _ of the electrons .
expanding terms in @xmath76 , we get , using the notation of figure [ dustscatdiag ] : @xmath77^{1/2 } \cdot r \cdot | \cos\psi |.\end{aligned}\ ] ] the complex optical constant for the dust grain relates @xmath39 and @xmath67 via @xmath78 . substituting in ( [ omegaeq ] ) we get @xmath79^{1/2 } \cdot r \cdot |\cos \psi|\ ] ] where @xmath80 .
the position vector @xmath81 takes on all values between @xmath82 , where @xmath34 is the radius of the dust , and @xmath83 .
replacing @xmath84 with its maximum value , @xmath34 , the requirement that @xmath85becomes @xmath86 ) is equivalent to equation ( [ rglimit ] ) for @xmath87 , so if the rg assumptions are met coherent scattering will occur for at least some scattering angles .
we can rewrite equation ( [ firstrewrite ] ) using some trigonometry and the already - stated assumption that @xmath88 to get a better limit on the scattering angle : @xmath89 in the x - ray regime @xmath90 .
thus @xmath91 will much larger than unity for normal interstellar dust grains and @xmath92 1 kev x - rays , so we can use the small angle approximation for @xmath93 .
waves scattered at angles where @xmath94 , when the rg assumptions are met , will have intensities proportional to @xmath95 , the number of electrons in the grain squared .
to summarize , the rg approximation requires that the dust neither reflects photons nor significantly affects their phases .
the induced scattering wave generated at the entrance to the grain must add in phase with one created at the exit .
if it does not , then coherency is lost , and the amplitude of the scattered wave never becomes proportional to the number of electrons squared in the entire dust volume squared , but to a fraction of that number .
furthermore , the combination of the rg approximation with the drude approximation assumes that absorption is negligible at all energies .
this is approximately valid when @xmath96kev , although there are astronomically abundant elements in dust with atomic energy levels above 1 kev are magnesium ( k edge , 1.294 kev ) , aluminum ( k edge , 1.550 kev ) , silicon ( k edge , 1.828 kev ) , and iron ( k edge , 7.083 kev ) . below 1 kev , however , there are many k and l shell edges which can absorb x - rays , an effect which will be considered in 3.2 .
the breakdown of the rg approximation and the effect of absorption require the use of full mie theory to correctly calculate the intensity of the scattered light .
mie theory calculates the exact solution for the scattering and absorption of light from a spherical particle ( van de hulst , 1957 ) . unlike the rayleigh - gans approximation
, it does not assume all the scattering sites will add coherently and as a result , mie theory can correctly calculate the scattered intensity for particles even when @xmath97 . in the limit where @xmath98 the mie solution recovers the result of the rayleigh - gans approximation .
another advantage is that absorption can be included easily , using measured values for the index of refraction .
however , although the mie solution is exact for light scattering from a sphere , unlike the rg approximation it is not expressible in a useful analytic form .
the mie solution requires careful numerical computation , especially for large values of the `` size parameter '' @xmath99 where the series solution requires summation over a large number of terms to converge .
we use the mie code developed by wiscombe ( 1979 , 1980 ) , which was written for atmospheric use but has been tested for size parameters up to 20,000 , sufficient for our problem .
the observed scattered intensity is obtained by integrating the intensity of the light scattered by a single dust particle to the observer , over the grain size and spatial distribution along the los ( see eq .
( [ gen_scat ] ) in the appendix ) .
these calculations can be greatly facilitated by pre - calculating tables of ( @xmath100 ) for a range of energies , dust sizes , and scattering angles .
we must point out , however , that even the mie solution is only of an approximate nature , since dust particles are neither spherical nor compositionally homogeneous .
we know that the first assumption is incorrect , since only non - spherical grains can polarize light , as grains have been observed to do ( _ e.g. _ mathis , 1990 ) .
the second assumption is particularly important , since the wavelength of a 1 kev photon is only 1.24 , substantially smaller than the radius of any dust grain .
thus , the x - ray s path through the grain will determine which dust grain constituents it interacts with , and so the x - ray scattering will be affected by the grain s exact morphology .
exploring this possibility in detail would require a discrete dipole approximation to the dust grain constituents ( _ e.g. _ draine , 1988 ) , which is beyond the scope of this paper .
the observed halo intensity can be calculated for any x - ray energy and dust grain size , using either rg approximation or the mie solution .
any variation in the result is due solely to the differential scattering cross section @xmath29 term . in figure
[ fig : q_02 ] we plot @xmath101 for 2 kev x - rays scattering from a single dust grain , as a function of the scattering angle @xmath102 .
the differential scattering cross section was divided by the geometrical cross section to show the relative effect of grain size on the scattering intensity .
figure [ fig : q_02](a ) presents the results for both a graphite or silicate dust grain of radius @xmath103 m and figure [ fig : q_02](b ) is similar but for grains of size @xmath104 m .
the mass densities of the two types of dust , graphite and silicate , were 2.2g@xmath105 and 3.3g@xmath105 respectively .
the optical constants were calculated using the drude approximation , ignoring absorption , both for the rg and the mie cases , so that the variation in the figures is due solely to the approximate nature of the rg solution . , for a 2 kev x - ray scattering from a @xmath103 m dust grain , for both graphite and silicate dust grains .
the solid lines show the rayleigh - gans approximation using the gaussian approximation ; the dashed lines show the equivalent mie calculation . in this energy and
@xmath102-range the two methods should give identical results ; however , use of the gaussian approximation leads to deviations at large angles .
( b ) same , for @xmath104 m dust grains.[fig : q_02],title="fig : " ] , for a 2 kev x - ray scattering from a @xmath103 m dust grain , for both graphite and silicate dust grains .
the solid lines show the rayleigh - gans approximation using the gaussian approximation ; the dashed lines show the equivalent mie calculation . in this energy and
@xmath102-range the two methods should give identical results ; however , use of the gaussian approximation leads to deviations at large angles .
( b ) same , for @xmath104 m dust grains.[fig : q_02],title="fig : " ] figure [ fig : q_02 ] shows that at 2 kev , the gaussian fit to the rayleigh - gans approximation is adequate to the task ; it deviates substantially from the mie solution only at large angles , after the scattering intensity has dropped by at least an order of magnitude .
the deviation itself is largely due to the gaussian approximation .
using the rg approximation with the spherical bessel function duplicates the ripples in the large - angle scattering of the @xmath104 m grains and the faster drop seen for the 0.1@xmath51 m grains . when the requirement that @xmath106 is not satified , the agreement between the rg and the mie solution becomes substantially worse .
figure [ fig : q_005 ] is identical to figure [ fig : q_02 ] , but for x - rays of energy 0.5 kev .
the `` rule of thumb '' suggests that in this case the rg approximation will hold for @xmath103 m radius dust , but fail for the @xmath104 m radius dust .
figure [ fig : q_005 ] confirms this , showing an acceptable match between the rg and the mie calculations for the smaller grains , but showing far more total scattering with the rg approximation than in the mie solution for the larger grain size . in this case , using the exact bessel function solution does not help .
the larger dust grain has more scattering sites , and the rg approximation assumes all the scattered waves all add coherently ; the mie solution shows that for large dust sizes this coherency assumption fails , and the scattered intensity therefore drops .
figure [ fig : q_005 ] suggests that the approximation fails abruptly , as might be predicted from the exponential nature of equation ( [ jacksoneq ] ) .
m dust grain .
the assumptions necessary for the rg approximation do not hold , and the results show that the differential scattering cross sections calculated using the mie and rg approximation disagree substantially at all scattering angles.[fig : q_005],title="fig : " ] m dust grain . the assumptions necessary for the rg approximation do not hold , and
the results show that the differential scattering cross sections calculated using the mie and rg approximation disagree substantially at all scattering angles.[fig : q_005],title="fig : " ] we have so far used the drude approximation to the optical constants , ignoring any damping , in both the rg approximation and the mie solution .
although the drude approximation works well above 1 kev , real dust grain constituents have many k and l shell electrons with energy levels that are in the range @xmath107kev , and these electrons will absorb soft x - rays .
the effect of absorption can be included by using the atomic scattering factor @xmath108 ( henke , 1981 ) as a modifier to the drude approximation : @xmath109 where @xmath110 is the atomic number of the element .
however , we simply use the measured optical constants which include the absorption effects .
for silicate particles ( \{mgfesi}o@xmath111 ) we used the optical constants of martin & rouleau ( 1991 ) , and for graphite particles we used rouleau & martin ( 1991 ) .
these values are a smooth extension of the draine & lee ( 1984 ) and laor & draine ( 1993 ) results into the fuv and x - ray regime . at x - ray energies
, there is no distinction between graphite and amorphous carbon , so this complication need not be considered .
we also considered composite dust particles as described in mathis ( 1996 ) ; for this case , we used the bruggeman rule ( bohren & huffman , 1983 ) to calculate the optical constants of the mixture .
figure [ fig : abs](a ) plots the observed halo intensity for 2 kev x - rays and a distribution of grain sizes , calculated three different ways .
we presented the observed halo , rather than showing @xmath29 for a single dust grain size as was shown in figures [ fig : q_02 ] and [ fig : q_005 ] , to show the magnitude of the observable differences between the methods . for 2 kev x - rays ,
these differences are clearly small .
the halo intensity , given by equation ( [ scat_eq ] ) , was found by integrating @xmath29 over an evenly distributed populates of dust particles ( @xmath112 ) characterized by a size distribution @xmath113 with @xmath114 .
the grains themselves consisted of a mix of silicate and graphite grains with dust to hydrogen mass ratios of 0.0057 and 0.0032 respectively .
we note that the choice of the upper limit to the size distribution , @xmath115 m , significantly affects the result . with this power law size distribution ,
most of the mass is in the largest grains .
in addition , the distribution of the largest dust particles are not well - determined by extinction measurements , since when the grain size is much larger than the wavelength the extinction becomes independent of wavelength .
x - ray halos are therefore a particularly useful and sensitive probe of the number and size of large dust grains . in the rg approximation
the intensity is proportional to the sixth power of the radius of the dust .
the mie solution also predicts larger dust grains scatter more than small grains , although not by as large a power ( see figures [ fig : q_005](a ) and ( b ) ) . in general , however , the source function of the scattered intensity is heavily weighted towards large dust grains . size distribution .
the results are shown using the rg approximation , the mie solution with the drude approximation to the index of refraction ( _ i.e. _ , without absorption ) and for the mie solution using measured optical constants.[fig : abs],title="fig : " ] size distribution .
the results are shown using the rg approximation , the mie solution with the drude approximation to the index of refraction ( _ i.e. _ , without absorption ) and for the mie solution using measured optical constants.[fig : abs],title="fig : " ] figure [ fig : abs ] shows the cumulative effects of the approximations used in calculating the observed halo intensity .
the simple rg approximation ( with the gaussian fit to the bessel function and the drude approximation to the optical constants ) predicts the largest intensity . the exact mie solution , using the drude approximation , reduces the intensity since not all the scattered waves add coherently .
the mie solution , using actual optical constants which include absorption , has the lowest intensity as some of the impinging x - rays are absorbed in the dust
. the effect of absorption can be much larger , as shown in figure [ fig : abs](b ) , which is identical to figure [ fig : abs](a ) but for 0.5 kev x - rays .
the difference between the rg and the mie solution for a realistic dust grain size distribution is quite noticeable , and figure [ fig : abs](b ) shows that absorption is very important at this energy .
we note in passing that not all of the absorbed x - ray s energy will be deposited in the dust ( dwek & smith , 1996 ) since the ejected photo - electron may deposit only a fraction of its energy in the solid .
we have shown that the accuracy of the rg approximation is size dependent and is in general less accurate at larger grain sizes .
the interstellar grain population is characterized by a mixtures of grain sizes and it is therefore interesting to apply mie theory to an astronomical situation . in figure
[ fig : mie_rg_comp ] , both the mie and rg methods were used to calculate the intensity of scattered x - rays from silicate grains as a function of x - ray energy , at the observed scattering angle @xmath116 .
we see that for higher energies , the shape of the curves for the two methods begins to agree when @xmath117kev , but below this energy there are substantial differences .
the two methods never agree exactly because of absorption , which is more important for larger grains but is not negligible even for the 0.1 @xmath51 m grain .
the presence of absorption lines in the grains can also be seen in the halo shape , which has irregular peaks and valleys in the mie case . for silicate grains of radius 0.1 and 0.4 @xmath51 m , calculated both by the rg approximation ( without absorption ) and the mie solution ( including absorption ) as a function of x - ray energy . for @xmath118kev
, the two methods show the same energy dependence ; below it , absorption modifies the @xmath119 dependence , an effect that becomes more pronounced at 0.4@xmath51 m .
[ fig : mie_rg_comp ] ] we conclude that in many astronomical situations using the rg approximation to calculate the halo intensity systematically overestimates the amount of x - ray scattering from a given dust size distribution .
consequently , the rg approximation can not be used for x - ray energies below 1 kev , except as an upper limit to the halo intensity . even for energies above 1 kev , rg is only an upper limit since absorption is never totally negligible , and will reduce the total intensity of scattered light .
the total scattered light intensity will in practice depend upon the spectrum of the source @xmath3 , the spatial distribution of dust grains @xmath9 , and the grain size distribution @xmath120 .
we can measure the x - ray spectrum @xmath3 , directly , and we will assume that dust grains are evenly distributed , so that @xmath121 .
the last parameter , the dust grain size distribution @xmath120 , is not well known , especially for larger size grains .
currently , it is measured by fitting the extinction of visible and uv starlight . we will present results using two different grain models : ( 1 ) a bare core silicate and graphite mix , with an @xmath122 power law size distribution ( following mathis , rumpl , & nordsieck , 1977)(mrn ) ; and ( 2 ) the recent composite fluffy dust model of mathis ( 1996)(m96 ) a distribution centered near 0.1 , consisting of composite silicate and graphite
we now apply the rg and mie solutions with these grain models to the observed scattering from the x - ray source ( nova cygni 1992 ) .
nova cygni 1992 exhibited a strong x - ray halo in the rosat pspc ( mathis _ et al .
_ , 1995 ) .
the observed spectrum was formally fit with a blackbody with temperature @xmath123ev , absorbed by an equivalent hydrogen column density of @xmath124@xmath125 .
( 1995 ) do not claim this is a true model of the nova , only that it is simple to calculate and fits the observed spectrum reasonably well .
the observed spectrum peaked at about 0.5 kev , and with this choice of assumed spectrum most of the photons from the source were absorbed .
however , this column density is not the best choice for use in the scattering intensity equation ( [ scat_eq ] ) , since the value of is highly dependent upon the underlying true spectrum .
in addition , the hydrogen column density measured from x - ray absorption depends upon the helium and metals along the los , and is not directly related to the number of dust grains along the los .
therefore , x - ray absorption measures of the hydrogen column density may not accurately reflect either the actual hydrogen column density or the dust grain column density .
the scattered halo intensity scales linearly with the column density of dust grains , which can be found by measuring the extinction along the los .
( 1995 ) ( and references therein ) estimated the reddening e(b - v ) to be between 0.19 - 0.31 mag . using the standard extinction law value of @xmath126 , this translated into a hydrogen column density in the range @xmath127@xmath125 .
there is a significant difference between the x - ray and extinction derived hydrogen column densities .
the former depends upon the source spectrum over a broad range of energies , whereas the latter depends on the source spectrum over a much narrower range ( b and v bands ) , and provides a more direct measurement of the dust column density along the los .
we therefore prefer the extinction determined hydrogen column density , and use a value of @xmath128@xmath125 . for the nova cygni source mathis _
( 1995 ) noted that this column density creates only a small optical depth for scattering , so the assumption of single - scattering is adequate .
nonetheless , the halo was clearly visible in the data to angles greater than @xmath129 before it merged with the background .
mathis _ et al . _
( 1995 ) fitted the halo using the rg approximation for the differential scattering cross section and found that an mrn model with a column density chosen to produce the observed visual extinction would scatter more x - rays than were seen . noting that the intensity of the scattered light is proportional to the grain density squared ,
they proposed a model consisting of composite fluffy dust ( cfd ) particles with substantial amounts of vacuum which would scatter fewer x - rays but would nevertheless produce the same observed visual extinction .
the amount of light scattered from a fluffy grain is less than that from the same size dense grain , since the fluffy grain has fewer electrons with which to scatter the x - rays .
this reduction in scattering remains even when the total amount of mass in dust is constant , since the intensity is proportional to the number of electrons in each grain squared .
therefore , having twice as many grains which are 50% vacuum will lead to a net 50% drop in the intensity when the rg approximation is valid .
in the 0.5 kev energy range , however , the rg approximation is not valid for dust grains larger than @xmath130 m , for two reasons .
first , the scattered waves will not add coherently as the dust grain material will affect their phases substantially , and secondly , absorption in the dust grain will remove some of the x - rays altogether . as a result , using the rg approximation with the no - damping drude approximation to the optical constants will tend to substantially overpredict the amount of scattered light and so the mathis _ et al . _ ( 1995 ) result must be re - evaluated using the mie solution . to examine the robustness of the conclusions derived by mathis _
( 1995 ) for the nova cygni data , we modeled the same data using mie theory for the standard mrn model , and the newly - proposed cfd model ( m96 ) .
the parameters of the two models are presented in table [ dustmodels ] .
the two dust models produce an overall equally good fit to the average interstellar extinction curve ( dwek , 1997 ) . however , the cfd model produces a visual extinction that is 6.7% higher than the mrn model , so we normalized the x - ray scattering intensity downward by the same amount . with this renormalization
the two models produce the same amount of visual extinction towards the x - ray source .
we limited our results to the case of evenly distributed dust , _
i.e. _ @xmath121 . .dust
grain parameters[dustmodels ] [ cols="<,<,^ , < , < , < , < " , ]
x - ray scattering is a powerful tool for measuring the composition and size distribution of interstellar dust . however , many approximations and assumptions are necessary to create models of the data .
we have assumed , for example , that dust grains are spherical , homogeneous , and evenly distributed throughout the ism , assumptions whose effects should be tested .
we have explored the use of the rayleigh - gans approximation for the differential scattering cross section , which when combined with the drude approximation to the index of refraction provides a very convenient analytical form for discussing x - ray scattering .
we find that the rg approximation works well for normal interstellar dust parameters and energies at or above 2 kev , but below this energy it overestimates the halo intensity .
we re - examined the nova cygni 1992 data , which has a strong x - ray halo , and found that it does not distinguish between fluffy and dense dust .
mie calculations using measured optical constants produce equally good fits for the standard mrn model , and for fluffy dust particles .
consequently the nova cygni data can not be used as evidence that interstellar dust particles are fluffy , though they may be .
the future of x - ray scattering is bright .
although the rosat pspc has the highest energy - resolved spatial resolution available currently , the axaf ccd imaging spectrometer ( acis ) and high - resolution camera ( hrc ) detectors will have arcsecond resolution with low to moderate energy resolution . clearly , nearly every point source seen with axaf will include an x - ray halo of some kind
. the high angular resolution will allow observers to extract halos down to @xmath131 or less , where the scattered intensity is due almost entirely to grains very near the source .
this will be a bonanza for dust grain theorists , as it will provide a new handle on the size distribution of larger grains as well as the position - dependence of grain populations . at the same time it will add a difficulty to the data analysis of diffuse sources , since a calculation of the x - ray halo due to dust scattering will have to made in order to remove it , in a process similar to the `` dereddening '' required of optical and uv measurements .
we thank alain lger for invoking our interest in this problem .
this work was performed while one of the authors ( rks ) held a national research council
gsfc research associateship .
this research was also supported by nasa astrophysics theory program grant # 344 - 02 - 05 - 01 .
consider a dust particle of size @xmath34 at a point p , illuminated by an x - ray source ( s ) a distance @xmath132 away , which scatters photons ( of energy @xmath119 ) towards an observer ( o ) at a distance @xmath133 , as is shown in figure [ fig : scat_diag ] . given a source luminosity of @xmath134 , the flux at p
is @xmath135 , where @xmath136 is the column density along the line sp and @xmath137 is the average interstellar absorption cross section at energy @xmath119 .
then the luminosity of scattered photons into solid angle @xmath138 is @xmath139 now consider the photons arriving at the observer , who measures the flux of photons from the scattered dust particle , using a detector of size @xmath140 , and a telescope with opening angle @xmath141 .
then we can choose @xmath138 so that @xmath142 , and the observer obtains a photon count rate ( with the opening angle @xmath141 for the telescope ) of @xmath143 this is the result for a single dust particle ; we can now integrate it over all dust particles in the scattering volume .
assume that there are @xmath144 dust grains per unit volume at position p. we will integrate over the line of sight using a differential volume element with length @xmath145 and solid angle @xmath141 .
the volume is @xmath146 , and we obtain the total count rate by integrating over @xmath133 : @xmath147\}}\over{4\pi r_1 ^ 2 } } \big({{d\sigma}\over{d\omega}}\big ) dr_2.\end{aligned}\ ] ] thus the count rate ( in photons per second ) in a detector with area @xmath140 and opening angle @xmath141 is given by @xmath148
. we can therefore write the intensity of the scattered light as @xmath149\sigma(e)\ } } \over{4\pi r_1 ^ 2 } } \big({{d\sigma}\over{d\omega}}\big ) dr_2.\ ] ] this equation must then be integrated over a energy bandpass and a range of dust size and position distributions . in general ,
the function @xmath150 will be a function of dust radius @xmath34 as well as a function of position .
therefore , the total scattering for a given energy range @xmath151 will be @xmath152\ } } \over{4\pi r_1 ^ 2 } } \big({{d\sigma}\over{d\omega}}\big ) dr_2\,da\,de.\ ] ] this equation is completely general ; however , we can simplify it substantially if the observed angle @xmath13 is very small , limiting the observations to very small angles . in this case
, the path traveled by a scattered photon will be very close to that traveled by an unscattered photon and the total hydrogen column density for the source , @xmath153 , will be nearly the same as that of the scattered photons , @xmath154 . as a result , we have the following approximation : @xmath155\ } } \over{4\pi r_1 ^ 2 } } \approx { { f_x(e ) \cos(\phi-\theta)}\over{(1-z)^2}}\ ] ] where @xmath156 is the observed flux from the source .
we can then rewrite equation ( [ gen_scat ] ) in terms of the normalized projected position coordinate @xmath25 and the scattering angles @xmath13 and @xmath102 : @xmath157 we can reduce this further by assuming that the dust size and position distribution is separable , and is independent of the hydrogen density @xmath158 , so that @xmath159 we have used the assumption of similar paths for the scattered and direct photons here as well . using these , and assuming @xmath160 , we can rewrite ( [ first_approx ] ) , to get @xmath161 where @xmath162 is the hydrogen column density to the source . the final assumption necessary to derive equation ( [ scat_eq ] ) in the text
is that @xmath163 .
we have already limited the observed angle @xmath13to small angles . although close to the x - ray source @xmath102 can be large even for very small values of @xmath13 , @xmath29
is strongly suppressed at scattering angles above @xmath164 , so that in practice , all the contribution to the scattered light comes from scattering angles below @xmath164 .
therefore the actual scattering angle @xmath102 will be small , so @xmath165 .
note also that @xmath166 for all geometries . for a nonuniform dust distribution ,
in which a significant amount of dust is near the source the contribution from larger scattering angles may be important , and an exact expression for @xmath167 will then be required . | small angle scatterings of x - rays by interstellar dust particles create halos around x - ray sources .
the halo intensity and its projected radial distribution around a source can provide important information on the spatial distribution of the dust along the line of sight to the source , and on the physical properties of the scattering dust particles .
halos around x - ray point sources have been used by several authors to infer that the scattering dust particles are fluffy in nature , consisting of aggregates of smaller refractory particles with vacuum occupying a significant fraction of their volume .
the nature and morphology of interstellar dust particles has recently gained new importance , since fluffy , composite dust particles have been suggested as a possible solution to the interstellar carbon `` crisis . ''
this crisis results from the discrepancy between the abundance of carbon in the interstellar medium available for creating dust , and the significantly larger amount of carbon that must be in dust in order to account for the uv - optical interstellar extinction in the diffuse ism .
previous studies of x - ray scattering have used the rayleigh - gans ( rg ) approximation to the differential scattering cross section for calculating halo properties .
however , the validity of the rg approximation fails for energies below 1 kev .
we use the exact mie solution for the differential scattering cross section and find that for these energies the scattering becomes much less efficient than is predicted by the rg approximation .
furthermore , the effects of k and l band absorption by atoms in the dust become important .
the net effect is that the rg approximation systematically and substantially overestimates the intensity of the halo below 1 kev , relative to the mie solution result .
in particular mathis _
et al . _ ( 1995 ) used the weaker - than - expected halo intensity observed around nova cygni 1992 to conclude that interstellar dust must be fluffy .
using the mie solution to the scattering intensity and including the effects of absorption , we find that , contrary to the conclusion of mathis _ et al . _ ( 1995 )
, the halo around nova cygni 1992 does not _ require _ interstellar dust grains to be fluffy in nature , and that the data is consistent with scattering from a mix of bare refractory silicate and carbon grains as well . |
we fix a prime @xmath6 , and an integer @xmath7 .
we use @xmath4 to denote mahowald s inverse limit construction @xmath8 , where @xmath9 stands for either @xmath10 or its analogue for @xmath11 when @xmath6 is odd ; see @xcite or @xcite .
this is an abbreviation for the fixed points of the greenlees may tate cohomology functor ; we write @xmath12 for what would be denoted in @xcite by @xmath13 . in particular
if @xmath14 is a ring spectrum , then so is @xmath12 .
the starting point of this paper is a calculation on coefficients , which is by now well known ( see lemma [ t - le : coefficients ] ) . if @xmath14 is a complex oriented ring spectrum in which the series @xmath15 ( x)$ ] is not a zero divisor in @xmath16 \ ! ] } = e^{*}\mathbf{c}p^{\infty } $ ] , then @xmath17(x)),\ ] ] where @xmath18 and @xmath19 .
here we use the awkward grading to emphasize that the ring is related to @xmath20 and @xmath21 . graded more conventionally
, we have @xmath22 ( x))\ ] ] where @xmath23 . throughout this paper
we will use @xmath24 to denote the ring of formal laurent series over @xmath25 that are allowed to be infinite series in @xmath26 , but only finite in @xmath27 . if @xmath14 is a ring spectrum and @xmath26 an indeterminate in degree @xmath28 , we write @xmath29 for the ring spectrum given by @xmath30 .
the multiplication is defined by the inverse limit of the obvious maps @xmath31 this gives @xmath32 = e _ { * } { ( \ ! ( x ) \ ! ) } $ ] .
( note that the theoretical possibility of phantom maps in ( [ ring - wedge ] ) means the multiplicative structure there may not be unique , but we are content to use whichever multiplicative structure might occur .
actually work of hovey and strickland shows that there are no phantom maps in this situation @xcite , so the multiplication is defined uniquely . )
note also that if @xmath14 is connective , @xmath33 in the succeeding , if we refer to a multiplication on @xmath34 it will be the one coming from the equivalence ( [ limprod ] ) , and the ring structure on @xmath35 for a connective graded ring @xmath25 will be understood to be the one given by the additive isomorphism with @xmath24 .
if @xmath36 is a flat module over @xmath37 , we write @xmath38 for the spectrum representing the homology theory @xmath39 this is a module spectrum over @xmath14 ( again , it will not matter for us that the module structure is well - defined only up to phantom maps ) .
we also use @xmath15 ( x)$ ] for the @xmath6series of the formal group law over @xmath14 given by the orientation .
when necessary we will decorate @xmath15 $ ] with a subscript indicating the formal group law .
we will work with a number of different spectra @xmath14 , all closely related .
the cohomology theory closest to @xmath40 is @xmath41 , the version of @xmath40 with singularities satisfying @xmath42 $ ] where @xmath43 is the @xmath44th araki generator .
the johnson
wilson theory @xmath45 has the obvious coefficient ring obtained from inverting @xmath46 we will also need to consider some simple variants of @xmath47 .
we list the theories below and their coefficient rings .
they are all flat over @xmath47 and are thus determined by their coefficients : @xmath48 \ ! ] } [ v_{n},v_{n}^{-1 } ] = { ( { e ( n)}_{*})^{\wedge}_{i_{n-1}}}\\ ( { { e_{n } } } ) _ { * } & = { e^{\wedge } ( n)}_{*}[u]/ ( u^{p^{n}-1}-v_{n})\\ & \text{\qquad\qquad\qquad and in later sections}\\ { { e ( n)}[{{w } } ] } _ { * } & = { e ( n)}_{*}[{{w}}]/ ( { { w}}^{p^{n-1}-1}-v_{n-1 } ) . \end{aligned}\ ] ] the `` @xmath49 '' is meant to suggest `` complete . ''
the reader should note that our choice of @xmath50 so that @xmath51 \ ! ] } [ u , u^{-1}]$ ] conflicts with the common convention that @xmath50 denote the same theory extended by the witt vectors of @xmath52 .
we make no use of that theory in this paper , so this should cause no confusion . the multiplicative structure on @xmath53 is given by the composite @xmath54 the other two spectra are finite extensions of @xmath53 , so the multiplicative structure is determined in the obvious way from the coefficients .
perhaps a few words of history are relevant here .
lin s theorem @xcite shows that for any finite cw
spectrum @xmath55 , the map @xmath56 is @xmath6completion .
this is not true @xcite for @xmath57 or various other @xmath40module spectra , though there is some predictable behavior , which is quite different from that of finite complexes . in particular
, @xmath58 is quite large in these cases . by a lovely but simple argument involving the thick subcategory theorem , mahowald and shick @xcite showed ( for @xmath59 )
that if @xmath55 is a type @xmath7 finite complex , and @xmath60 is a @xmath61self map , then @xmath62 .
this is the starting point for a series of observations .
for example , it turns out that @xmath63 ; indeed @xcite proves that @xmath64 for any finite group @xmath65 .
hopkins has shown that if @xmath55 is of type @xmath7 then @xmath66 if @xmath55 is type @xmath7 ( this is proved in @xcite ) .
calculations of hopkins and mahowald ( including a proof when @xmath67 ) lead them to conjecture that if @xmath55 is type @xmath68 , then @xmath69 this is related to the chromatic splitting conjecture @xcite . in light of ( [ hopkins - mahowald ] ) , it seems worthwhile to investigate @xmath2 . in @xcite
it is shown that @xmath2 is @xmath70periodic in an appropriate sense . using this
, one sees in @xcite that when @xmath55 is finite , the _ bousfield class _ of @xmath71 is compatible with the conjecture ( [ hopkins - mahowald ] ) ( for @xmath59 , this is also proved by different techniques in @xcite ) .
the goal of the present paper is to give as precise a description as possible of @xmath2 in terms of familiar @xmath70periodic spectra like @xmath3 .
we hope that with the results of this paper in hand , it may be possible to make further headway on ( [ hopkins - mahowald ] ) .
the first step , in section [ sec : coefficients ] , is a series of calculations of @xmath72 in terms of more familiar objects for various @xmath14 .
it turns out to be useful to complete with respect to @xmath73 .
since @xmath2 is already @xmath3 local , this amounts to localizing with respect to @xmath74 .
this completion is in some sense not very dramatic ; in particular if @xmath55 is type @xmath68 as in ( [ hopkins - mahowald ] ) , the completion leaves @xmath71 unchanged .
the main result , for the case of @xmath47 , is : [ it : en - coeffs - iso - equation ] there is an isomorphism of rings @xmath75 section [ sec : module - spectrum ] gives splittings of the spectra @xmath76 and @xmath77 in the manner suggested by theorem [ it : en - coeffs - iso - equation ] .
for @xmath77 the result is : [ it : tate - splitting ] there is a map of spectra @xmath78 which after completion at @xmath79 ( or equivalently after localization with respect to @xmath74 ) induces the isomorphism of theorem [ it : en - coeffs - iso - equation ] on homotopy groups .
we emphasize that the map in theorem [ it : tate - splitting ] is not multiplicative , even though the left hand side can be given a multiplication as in equation ( [ ring - wedge ] ) and the right hand side is a ring spectrum by @xcite , and theorem [ it : en - coeffs - iso - equation ] gives an isomorphism of rings . now
the homotopy calculation in section [ sec : coefficients ] also shows that the obvious formal group law over @xmath77 has height @xmath68 , so one might view the map @xmath80 as a sort of chern character , the classical chern character being the case @xmath67 . to make sense of this ,
one ought to use the calculations of section [ sec : coefficients ] to construct a map of ring spectra . to do this
, we give an isomorphism between the natural formal group law over @xmath2 and the honda formal group law of height @xmath68 .
this can be done after a dramatic base extension , but in section [ sec : isomorphism ] we show that it can then be done in a canonical way .
section [ sec : generalized - chern - character ] uses section [ sec : isomorphism ] and the lubin
tate theory of lifts to construct an isomorphism of ring spectra between an extended version of @xmath2 and an extended version of @xmath3 .
[ iso - ring - spec - theorem ] these is a canonical equivalence of complex oriented ring spectra @xmath81 the objects discussed in theorem [ iso - ring - spec - theorem ] are defined precisely in sections [ sec : isomorphism ] and [ sec : generalized - chern - character ] .
briefly , the ring @xmath82 is isomorphic to the @xmath6adic completion of @xmath83 ( see lemma [ t - le : cohenofres ] ) .
the ring @xmath84 is a complete discrete valuation ring with maximal ideal @xmath85 and residue field @xmath86 .
@xmath87 is a finite extension of @xmath2 completed at @xmath79 and defined in equation ( [ bigtate ] ) , and @xmath88 is defined as usual to be a completed @xmath89periodic extension of @xmath90 as usual so that ( [ def : gltjw ] ) holds .
the authors thank johns hopkins university , at which the conversations leading to this paper took place , and max - planck - institut fr mathematik , at which the last fragments were committed to paper .
the authors were partially supported by the nsf .
we assume that @xmath14 is complex oriented , and that @xmath15 ( x)$ ] is not a @xmath91 divisor in @xmath16 \ ! ] } $ ] . in @xcite
it is shown that @xmath12 is an inverse limit of @xmath14 smashed with thom spectra over @xmath11 .
in particular , @xcite shows that @xmath92 ( x))$ ] as a module over @xmath16 \ ! ] } / ( [ p ] ( x ) ) = e^ { * } ( b\mathbf{z}/ ( p)_{+})$ ] .
now @xcite proves that the map @xmath93 is a map of rings .
it follows that the element `` @xmath27 '' in @xmath92 ( x))$ ] satisfies @xmath94 in the ring @xmath72 , so really is the inverse of @xmath26 . from this one
concludes that [ t - le : coefficients ] there is an isomorphism of rings @xmath95(x)).\ ] ] recall @xmath24 is @xmath96 \ ! ] } $ ] where @xmath97 is the multiplicatively closed subset generated by @xmath26 .
if @xmath25 is graded connected , and @xmath26 has degree @xmath98 , then additively @xmath99 . as remarked in the introduction , wherever this notation occurs , we will also consider @xmath100 to have the ring structure given by this isomorphism .
similarly , when @xmath14 is a connective ring spectrum , we will consider the ring structure on @xmath101 as in ( [ ring - wedge ] ) .
the following observation motivates conjecture 1.6 of @xcite ( which was corrected as conjecture 1.2 of @xcite ) .
[ t - pr : bp < n>-coeff - iso ] @xmath102(x ) ) \nonumber \\ & \simeq & \prod_{k \in { \mathbf{z } } } \sigma ^{2k}{{bp\langle n-1 \rangle } _ { p}\!\!\!\hbox{\^ { } } } _ { * } \\ & \simeq & { { bp\langle n-1 \rangle}_{p}\!\!\!\hbox{\^ { } } } _ { * } { ( \ ! ( x ) \ ! ) } .
\nonumber \end{aligned}\ ] ] the first line is equation ( [ coefficient - calculation ] ) . we wish to simplify the right hand side . for convenience ,
we introduce degree @xmath91 elements @xmath103 then the ideal in the quotient above is generated by the relation @xmath104 where @xmath105 is the formal group law over @xmath106 induced by the orientation from @xmath40 .
expanding the formal sums and dividing by @xmath26 , we can write the relation in the form @xmath107 and this equation implies @xmath108 by iterating this relation some finite number of times ( depending on the multi - index @xmath109 ) , we produce a polynomial @xmath110 so that @xmath111/ ( w_{1}^{i_{1}},\dots , w_{n-1}^{i_{n-1}}).\ ] ] we thus produce a power series @xmath112 $ ] so that @xmath113 \subseteq { \bf z}_{p}[v_{1},\ldots , v_{n-1 } ] { [ \ ! [ x ] \ ! ] } .\ ] ] thus the map @xmath114 which sends @xmath43 to @xmath43 ( @xmath115 ) , @xmath26 to @xmath26 and @xmath61 to @xmath116 is a well defined ring map .
there is a map @xmath117 defined in the obvious way , that becomes an inverse map upon extending it to @xmath118 .
this extension exists , because the relation @xmath119 ( x ) = 0\ ] ] together with araki s formula for the @xmath6series ( * ? ? ? *
a2.2.4 ) allows us to write @xmath120 ( v_{1}x^{p}+_{f}\dots + _ { f}v_{n}x^{p^{n}})\ ] ] where @xmath105 is the formal group law on @xmath41 induced by the complex orientation from @xmath40 .
now dividing ( [ formula - for - p ] ) by @xmath26 lets us to write any @xmath6adic integer as a power series in @xmath26 .
the conjecture of @xcite is that the isomorphism is the effect on homotopy groups of an equivalence of spectra @xmath121 this is proved for @xmath67 in @xcite and for @xmath122 in @xcite using the adams spectral sequence .
of course in general an isomorphism in homotopy does nt give an isomorphism of spectra , but it does if it is an isomorphism of @xmath123modules satisfying landweber exactness .
this suggests inverting @xmath61 to produce an isomorphism of spectra .
we shall show that there is an isomorphism of the expected form after an appropriate completion , taking as our starting point the spectra @xmath124 and @xmath2 .
the first point is that these spectra are not the same . the map @xmath125 gives a map @xmath126 , and since @xmath61 is a unit on the right , this gives a map @xmath127 .
on homotopy we get @xmath128(x))]\\ & \hookrightarrow & { e ( n ) } _ { * } { ( \ ! ( x ) \ ! ) } /([p](x))\\ & = & ( t{e ( n ) } ) _ { * } = \pi _ { * } t ( v_{n}^{-1}bp\langle n \rangle).\end{aligned}\ ] ] to see that the inclusion is proper , set @xmath129 and notice that the series @xmath130 is an element of @xmath131(x))$ ] but not of @xmath132(x))$ ] .
this reflects the fact that @xmath4 need not commute with direct limits ( nor in fact does it generally commute with inverse limits ) .
we treat the case @xmath133 first .
we would like to extend the isomorphism of ( [ bp < n>-iso ] ) to an isomorphism @xmath134 for the map to exist we need the image of @xmath135 under the isomorphism in ( [ bp < n>-iso ] ) to be a unit after inverting @xmath70 on the range of that map .
this is false for @xmath136 .
we can see this by checking that the image of @xmath135 after inverting @xmath70 is not a unit even modulo @xmath137 . the range of ( [ vbp < n>-iso ] ) in dimension @xmath91 , modulo @xmath137 , is @xmath138 \ ! ] } [ w_{n-1}^{-1}]$ ] .
we have @xmath139 , so @xmath140 \ ! ] } } .\ ] ] since @xmath141 is a unit in the range , the image of @xmath135 is a unit if and only if the image of @xmath135 divided by @xmath141 , which we will abuse notation by writing @xmath142 , is a unit . continuing to work modulo
@xmath143 , @xmath144 \ ! ] } } .\ ] ] now if we examine the inclusion of rings @xmath145 \ ! ] } [ w_{n-1}^{-1 } ] \subseteq { ( { { \mathbf{z}}_{p } } { [ \ ! [ w_{n-1 } ] \ ! ] } [ w_{n-1}^{-1}])_{p}\!\!\!\hbox{\^ { } } } , \ ] ] we see that since @xmath146 , @xmath142 is a unit in the completion , with inverse given by @xmath147 .
if we write this element as a power series in @xmath141 plus a power series in @xmath148 , we see that @xmath149 is the sum of a power series in @xmath141 with @xmath150 where @xmath151 .
the power series in @xmath141 is neccessarily in the image of @xmath152 , but @xmath153 is clearly not .
it follows that @xmath149 is not in the image of @xmath154 , and hence @xmath142 is not a unit in @xmath138 \ ! ] } [ w_{n-1}^{-1}]$ ] .
similarly , if the map existed , it could be an isomorphism only if @xmath70 were a unit in the domain after inverting @xmath61 .
this also is not the case for @xmath155 , for similar reasons . when @xmath67 both of these conditions are met , the map above exists , and is an isomorphism .
[ funny - iso ] if the map from ( [ bp < n>-iso ] ) is completed at the ideal @xmath156 then there is an isomorphism @xmath157)^{\wedge}_{i_{n-1}}}.\ ] ] we follow the isomorphism of ( [ bp < n>-iso ] ) @xmath158(x ) ) \\ & \simeq & { { bp\langle n-1 \rangle}_{p}\!\!\!\hbox{\^ { } } } _ { * } { ( \ ! ( x ) \ ! ) } \end{aligned}\ ] ] by the inclusion @xmath159)^{\wedge}_{i_{n-1}}}\ ] ] into the module obtained by inverting @xmath70 and then completing .
now note that @xmath135 has image as in ( [ wn - image ] ) .
so dividing by @xmath160 we see the image of @xmath135 divided by @xmath141 is @xmath161 plus terms in the ideal @xmath79 and terms in @xmath162 \ ! ] } $ ] .
this is a unit , so since we have inverted @xmath70 , the image of @xmath135 is a unit .
this allows us to extend our map to the domain given by inverting @xmath61 .
since the range is complete , we can also extend to the completion .
on the other hand , a similar argument allows us construct the inverse map from the inverse map of ( [ bp < n>-iso ] ) .
now we shall show that the isomorphism in ( [ comp - iso ] ) is induced by a map of spectra @xmath163 both sides are ring spectra with obvious @xmath164module structures
. they would be isomorphic as @xmath164algebras by the landweber exact functor theorem if we could make them @xmath164algebras so that the coefficient isomorphism preserves the map from @xmath123 .
let @xmath25 be the ring spectrum on the right .
in order to construct a map of ring spectra inducing the isomorphism in ( [ comp - iso ] ) , it is necessary that the fgl induced by @xmath165 be isomorphic to the `` usual '' fgl on @xmath166 ( induced by @xmath167 ) .
we can not demonstrate such an isomorphism , but we can exhibit an isomorphism of spectra ( that preserves neither the @xmath164module structure , nor the multiplicative structure ) .
we do this in section [ sec : module - spectrum ] .
the situation for @xmath77 is very similar . we can attempt to construct a map @xmath168 as in ( [ bp < n>-iso ] ) , but we immediately run into the problem that @xmath135 , and hence @xmath61 , does not go to a unit for @xmath155 . also as before , @xmath70 is not a unit on the left .
the solution is the same ; after completing both sides at @xmath79 we can construct an isomorphism .
[ en - coeff - iso ] the map of ( [ bp < n>-iso ] ) extends to an isomorphism @xmath169 ( x)^{\wedge}_{i_{n-1 } } } = { { e ( n-1 ) } _ { * } { ( \ ! ( x ) \ ! ) } ^{\wedge}_{i_{n-1}}}\ ] ] where @xmath170 ( x)^{\wedge}_{i_{n-1}}}\ ] ] and @xmath171 in the next section , we use this calculation to construct an isomorphism of spectra .
we are able to do this without showing the corresponding formal group laws are isomorphic , so we do not get an isomorphism of @xmath164modules . in section [ sec : generalized - chern - character ] , we extend scalars suitably to construct an isomorphism of formal group laws , yielding isomorphisms of @xmath164algebra spectra
we begin with an algebraic observation . [ algebraic - splitting ] let @xmath174 be a complete local ring , @xmath175 .
let @xmath36 be an @xmath176module such that the map @xmath177 induces isomorphisms @xmath178 let @xmath179 be an index set for a vector space basis of @xmath180 , so that @xmath181
. then there is a map @xmath182 which is an isomorphism when completed at @xmath183 .
we construct a map as follows .
let @xmath184 be a basis for @xmath180 as a @xmath185vector space .
let @xmath186 be lifts to @xmath36 .
then we map @xmath187 to @xmath36 by @xmath188 .
our hypotheses imply that this map gives an isomorphism on the associated graded with respect to the filtration induced by powers of @xmath183 .
this implies that the completion of the map is an isomorphism .
note that if the index set @xmath179 in lemma [ algebraic - splitting ] is infinite , @xmath36 will not generally be isomorphic to a free @xmath176module .
for example if @xmath189 and @xmath190 then @xmath36 is not a free @xmath191module . to see this , observe that no free @xmath191module of infinite rank can be @xmath6complete .
next we recall the result proved in ( * ? ? ?
* theorem 4.1 ) .
the identification @xmath192 $ ] gives a splitting @xmath193 where @xmath25 ranges over multi - indices of non - negative integers ( with only finitely many positive coordinates ) @xmath194 , @xmath195 and @xmath196 to build a map from right to left of ( [ bp - splitting ] ) , take @xmath197 to @xmath198 by using the homotopy class @xmath199 , smashing on the left with @xmath40 and then multiplying the left pair of @xmath40 s : @xmath200 the wedge over all @xmath25 of these maps is an isomorphism on homotopy groups , so is invertible , and @xmath201 is that inverse .
theorem 4.1 of @xcite states that the composite @xmath202 is a homotopy equivalence after smashing with a type @xmath203 spectrum and inverting @xmath204 .
the map @xmath205 is induced by leaving out wedge summands , and @xmath206 by the usual reduction of ring spectra @xmath207 .
we use the notation of @xcite , derived from @xcite : @xmath208 is the set of multi - indices with the first @xmath209 indices @xmath91 , and @xmath210 . we shall use the following facts about @xmath164module spectra @xmath36 : @xmath211 when @xmath212 is a finite type @xmath203 spectrum ( which follows from ( * ? ? ?
* theorem 1 ) ) ; @xmath213 where @xmath212 runs through finite type @xmath203 spectra under @xmath214 .
use of the nilpotence theorem , @xcite , is required to produce a sufficient supply of such @xmath212 and to ensure there are enough maps between them , as in ( * ? ? ?
* proposition 3.7 ) . for our purposes equation ( [ iadic - completion ] ) can be taken to be a definition , but see the remark below .
inverting @xmath204 , we get @xmath215 this last equality is by equation ( [ vn - limit ] ) combined with , say , the proof of ( * ? ? ?
* proposition 7.4 ) which verifies that @xmath216 as @xmath212 runs over finite type @xmath7 spectra under @xmath214 .
so if @xmath204 is already a unit on @xmath36 then @xmath217 and we will generally use the first notation rather than the second below .
although we use ( [ iadic - completion ] ) as the definition of @xmath218 , there are other approaches that can be taken for specific @xmath36 . for certain @xmath164modules
@xmath36 , one can define a spectrum @xmath219 with homotopy groups @xmath220 by using landweber exactness . in those cases , also by using landweber exactness , one can prove that @xmath221 as given in ( [ iadic - completion ] ) .
if @xmath36 has enough structure , one may also be able to define a completion of @xmath36 using either baas sullivan bordism with singularities @xcite , or structured ring spectra using the techniques of @xcite . in either case , one can use the nilpotence theorem to verify that the construction is homotopy equivalent to the one in ( [ iadic - completion ] ) .
each map in ( [ long - map ] ) except @xmath222 is a map of left @xmath40modules .
recall @xmath223 is the ideal @xmath224 . since @xmath223 is invariant , @xmath225 , and thus each map in ( [ long - map ] ) is compatible on homotopy with the @xmath223adic filtration .
let @xmath212 be a finite type @xmath203 spectrum ; then smashing ( [ long - map ] ) with @xmath226 gives an equivalence , and thus an @xmath227module structure on @xmath228 or , by taking inverse limits , a ( possibly not unique ) module structure on @xmath229 .
we prove the following proposition as a warm - up to our additive splitting of the tate cohomology spectrum .
the construction is a general method for splitting @xmath230 when @xmath105 is a nice enough @xmath40module , given the splitting of @xmath231 .
a theorem equivalent to proposition [ ej - splitting ] is proved as ( * ? ? ?
* theorem 4.7 ) .
[ ej - splitting ] if @xmath232 , there is a map @xmath233 which gives an equivalence after completing with respect to @xmath234 , or equivalently , after localizing with respect to @xmath235 .
the index @xmath236 runs through the monomials in @xmath237.\ ] ] @xmath238 is an @xmath231module , and thus an @xmath239module .
let @xmath240 . since @xmath241 where @xmath212 runs through finite type @xmath203 spectra , @xmath242 \ ! ] } [ v_{j}^{\pm 1},v_{j+1},\dots , v_{n-1},v_{n}^{\pm 1}].\ ] ] now since @xmath183 is an invariant ideal @xmath243 is well - defined , whether we think of @xmath183 as an ideal in @xmath244 acting on @xmath245 via the @xmath246module action on @xmath231 or as an ideal in @xmath247 acting via the associated localization map to @xmath248 .
we calculate that the associated graded to the @xmath183adic filtration on @xmath36 is @xmath249 \ ! ] } [ v_{j}^{\pm 1}]\ ] ] where @xmath236 runs through monomials in @xmath250 $ ] .
@xmath36 satisfies the hypotheses of lemma [ algebraic - splitting ] ( for the left @xmath183structure which comes from the map @xmath251 ) , where @xmath252 \ ! ] } [ v_{j}^{\pm 1}].\ ] ] now the splitting of @xmath231 gives @xmath238 an @xmath253structure , and thus an associated @xmath176action ;
the @xmath183adic filtration is the same , so @xmath36 also satisfies the hypotheses of lemma [ algebraic - splitting ] for that @xmath183adic filtration .
we can now make the usual homotopy theoretic argument : take generators of @xmath180 and lift them to elements of @xmath36 . use the @xmath246 structure of @xmath238 to make maps @xmath254 realizing these generators on the unit of the ring spectrum @xmath246 .
this gives a map @xmath255 by lemma [ algebraic - splitting ] , this map induces an isomorphism on homotopy groups after completion with respect to @xmath256 , ie after applying @xmath257 , which leaves the right hand side unchanged .
we apply techniques similar to those used in proposition [ ej - splitting ] to prove the following .
[ tate - splitting ] there is a map of spectra @xmath258 that becomes an isomorphism on homotopy groups after completion at @xmath79 or equivalently after localization with respect to @xmath74 .
we proceed as in the proof of proposition [ ej - splitting ] .
we have given @xmath259 as a @xmath247module in proposition [ en - coeff - iso ] .
it satisfies the hypotheses of lemma [ algebraic - splitting ] with @xmath260 , @xmath261 . since @xmath262
, we have an @xmath263action and hence an @xmath3action .
as above , the two available @xmath183adic filtrations on @xmath264 ( one from @xmath247 , the other from @xmath265 via the @xmath266structure on @xmath267 ) are the same since @xmath183 is an invariant ideal .
we now proceed in a slightly different manner .
note that @xmath268 for some indexing set @xmath179 , since @xmath269 is a graded field .
we could apply lemma [ algebraic - splitting ] and proceed as before , but we would actually like better control over our expression for @xmath270 . in particular , the index set @xmath179 in equation ( [ vector - space ] ) must be uncountable , but we would like to find a countable set of topological generators for @xmath270 .
in fact , we would like these generators to correspond to the ( positive and negative ) powers of @xmath26 . to accomplish this ,
we first recall ( * ? ? ?
* theorem 16.1 ) which states that @xmath271}$ ] , where @xmath272 is the usual complex line bundle .
recall also that the thom class of @xmath273 is in dimension @xmath274 , and is not torsion .
in fact the spectrum @xmath273 has a cw structure with exactly one cell in each dimension greater than or equal to @xmath274 .
since the @xmath274cell generates a non - torsion element of homology , the attaching map of the @xmath275 cell to the @xmath274 skeleton is null .
so the cell in dimension @xmath275 is spherical , and the inclusion of that @xmath275 cell , smashed with the unit of @xmath14 when @xmath14 is a ring spectrum , gives @xmath276.\ ] ] now , we take @xmath277 , and use the @xmath278structure to construct a sequence of maps @xmath279 we make a map @xmath280 by composing the map of ( [ partial - wedge - mapped - in ] ) with the map @xmath281 given by ( * ? ? ?
* theorem 16.1 ) .
taking inverse limits of the maps @xmath280 gives a map @xmath282 this map defines an isomorphism on the associated graded modules with respect to @xmath183 .
it follows that @xmath283 is an equivalence after completion , that is @xmath284^{\wedge}_{i_{n-1 } } } = { t{e ( n)}^{\wedge}_{i_{n-1}}}.\ ] ] -1truecm by a very similar argument one can prove the analog to proposition [ funny - iso ] : [ funny - equiv ] there is an equivalence of spectra @xmath285{^{\wedge}_{i_{n-1}}}\rightarrow ( v_{n}^{-1 } t{bp\langle n \rangle}){^{\wedge}_{i_{n-1}}}.\ ] ] we leave the proof to the interested reader . given all the completions that occur in this section and in section [ sec : coefficients ] , one might hope that by using some other , already complete theory like @xmath53 or @xmath286 , we could prove a theorem with a simpler statement .
this is unfortunately not the case .
there are similar results for these theories , but even if @xmath14 is complete with respect to @xmath287 , @xmath12 will generally not be , and so will need to be completed again .
there are variants of theorem [ tate - splitting ] for these other spectra as well , but the statement is not simpler .
in this section we shall take @xmath6 to be an odd prime and @xmath289 to be an integer .
it will ease the superabundance of superscripts to use the abbreviation @xmath290 .
we define a number of formal group laws in this section that are used in the remainder of the paper . for reference , we list them here .
1 . @xmath291 : a homogeneous formal group law of degree @xmath89 on @xmath292 induced by the usual orientation of @xmath1 .
@xmath293 : a twist of the pushforward of @xmath291 to @xmath294 by the element @xmath295 .
this is homogeneous of degree @xmath91 .
@xmath296 : a twist of the pushforward of @xmath291 to @xmath297}_{*}$ ] of @xmath291 by @xmath298 .
this is also homogeneous of degree @xmath91 .
@xmath299 : the pushforward of @xmath296 to the residue field of @xmath300 ; @xmath301 is defined in equation ( [ bigtate ] ) .
@xmath302 : the honda formal group law of height @xmath68 over @xmath303
@xmath304 : a formal group law introduced in the proof of proposition [ t - pr : e - phi - is - h ] that is shown in that proof to be the same as @xmath302 .
we use the canonical orientation of @xmath1 , which provides a coordinate so that @xmath305 \ ! ] } , \ ] ] and as usual , if @xmath306 is the multiplication on @xmath307 , @xmath308 \ ! ] } = e ( n)^ { * } { [ \ ! [ x ] \ ! ] } \hat{{\underset{}{\otimes } } } e ( n)^ { * } { [ \ ! [ y ] \ ! ] } = e ( n)^ { * } ( \mathbf{c}p^{\infty } \times \mathbf{c}p^{\infty } ) \ ] ] is a formal group law which we will denote @xmath309 .
this formal group law has the feature that its @xmath6series is given by @xmath310 recall that @xmath311 \ ! ] } [ u , u^{-1}]\ ] ] with @xmath312 and @xmath313 .
there is an isomorphism @xmath314 \ ! ] } \ ] ] in terms of which the coproduct on @xmath315 is determined by the formula @xmath316 where @xmath293 is the group law @xmath317 over @xmath318 .
since @xmath70 is a unit in @xmath319 , we shall also consider the theory @xmath297}$ ] obtained by adjoining an element @xmath320 of degree @xmath89 such that @xmath321 and then completing with respect to the ideal @xmath322 .
we prefer this choice to the usual parameter @xmath323 ( which gives @xmath50 ) because the functor @xmath4 will emphasize height @xmath68 behavior instead of height @xmath7 , and the normalization we choose to make things @xmath89periodic leads to simpler statements in section [ sec : isomorphism ] and this section .
equation shows that there is an isomorphism @xmath324}\cong \pi_{*}{{e ( n)}[{{w } } ] } { ( \ ! ( { { x } } ) \ ! ) } /{[p]}_{{g ( n ) } } ( { { x}}).\ ] ] proposition [ en - coeff - iso ] shows that @xmath70 , and so also @xmath320 , is a unit in the homotopy of the @xmath79adic completion @xmath325}\bigr){^{\wedge}_{i_{n-1}}}\ ] ] which is thus @xmath89periodic . if @xmath296 denotes the formal group law @xmath326 ( which implies @xmath15_{f } ( s ) = w[p]_{g ( n ) } ( w^{-1}s)$ ] ) and we introduce elements @xmath327 of degree zero , then the argument of proposition [ en - coeff - iso ] shows that there are isomorphisms @xmath328 { ( \ ! ( { { y } } ) \ ! ) } / ( { [ p]}_{{f } } ( { { y}}))\bigr]{^{\wedge}_{i_{n-1}}}\\ & \cong \label{eq : e - coefficients } \bigl[{{\mathbf{z}}_{p}}[{{w}}_{1},\dots , { { w}}_{n-2 } ] { ( \ ! ( y ) \ ! ) } \bigr]{^{\wedge}_{i_{n-1}}}\\ \pi_{*}{t e } & \cong \bigl[{{\mathbf{z}}_{p}}[{{w}}_{1},\dots , { { w}}_{n-2 } ] { ( \ ! ( y ) \ ! ) } \bigr]{^{\wedge}_{i_{n-1}}}[{{w}},{{w}}^{-1 } ] .
\notag\end{aligned}\ ] ] the group law @xmath296 is defined over @xmath329 $ ] and hence over @xmath330 ; it is @xmath6typical , and its @xmath6series satisfies the functional equation @xmath331_{{f } } ( t ) = pt { \underset{{f}}{+ } } { { w}}_{1}t^{p } { \underset{{f}}{+ } } \dots { \underset{{f}}{+ } } { { w}}_{n-2}t^{p^{n-2 } } { \underset{{f}}{+ } } t^{q } { \underset{{f}}{+ } } { { w}}_{n } t^{pq}.\ ] ] we denote by @xmath299 the image of the group law @xmath296 in the residue field of @xmath332 . [ t - pr : p - series - residue - field ] the residue field of @xmath330 is @xmath333 .
the element @xmath334 of @xmath330 maps to @xmath335 in the residue field .
the formal group law @xmath299 has coefficients in the subring @xmath336 $ ] , and its @xmath6series satisfies the functional equation @xmath337_{{{f}_{0 } } } ( t ) = t^{q}{\underset{{{f}_{0}}}{- } } { y}^{(1-p)q } t^{pq}.\ ] ] the statement about the residue field follows from equation . before we proceed further ,
recall that since @xmath299 is @xmath6typical and @xmath6 is odd , @xmath338_{{{f}_{0 } } } ( t ) = -t$ ] .
note that the image of equation in the residue field is @xmath119_{{{f}_{0 } } } ( t ) = t^{q } { \underset{{{f}_{0}}}{+ } } { { w}}_{n } t^{pq},\ ] ] so equation follows from the assertion that @xmath334 maps to @xmath335 . since @xmath15_{{{f}_{0 } } } ( { { y}})=0 $ ] we have @xmath339_{{{f}_{0 } } } ( { { w}}_{n } y^{pq } ) = - { { w}}_{n}{{y}}^{pq } \\ { { w}}_{n } & = -{{y}}^{(1-p)q}.\end{aligned}\ ] ] finally , @xmath299 is defined over the subring @xmath340 $ ] since @xmath296 is actually defined over the polynomial ring in @xmath341 , and @xmath342 is negative .
so @xmath299 is a @xmath6typical formal group law over @xmath340 $ ] , of height @xmath68 in the field @xmath343 or @xmath344 . on the other hand ,
let @xmath345 be the honda law of height @xmath68 over @xmath303 , characterized by the fact that it is @xmath6typical with @xmath6series @xmath119_{h } ( t ) = t^{q}.\ ] ] comparison with equation shows that @xmath346 now lazard @xcite proves that over the separable closure @xmath347 , there is an isomorphism of formal group laws @xmath348 which is in general not at all canonical .
in this section we show that there is a unique isomorphism @xmath349 which preserves equation in a suitable sense . to make this precise , note that expanding a rational function as a power series at infinity gives a map of fields @xmath350 which extends to @xmath351 [ t - th : jacks - psi ] there is a unique isomorphism @xmath352 of formal group laws over @xmath347 such that the image of @xmath353 in @xmath354 has coefficients in the image of @xmath355 \ ! ] } \rightarrow { { \mathbf{f}_{p } } { ( \ ! ( { { y}}^{-1 } ) \ ! ) } ^{\textup{sep}}}.\ ] ] @xmath353 gives an isomorphism over @xmath356 \ ! ] } $ ] satisfying @xmath357 we build up to the proof gradually ; the proof itself appears after proposition [ t - pr : e - phi - is - h ] .
[ t - le : tau ] there is a unique series @xmath358 { [ \ ! [ t ] \ ! ] } $ ] such that @xmath119_{{{f}_{0 } } } = [ p]_{{{h}}}\circ \tau.\ ] ] this power series has the properties that @xmath359 the equation which @xmath360 must satisfy is @xmath119_{{{f}_{0 } } } ( t ) = \tau ( t)^{q}.\ ] ] if @xmath361 then the functional equation becomes @xmath362_{{{f}_{0 } } } ( t ) & = \sum_{i , j } b_{ij } ( t^{q})^{i } ( -{{y}}^{(1-p)q}t^{pq})^{j } \\ & = \sum_{i , j } b_{ij } ( -{{y}}^{1-p})^{jq } t^{(i+pj)q}.\end{aligned}\ ] ] so we must show that @xmath363 has a unique @xmath364root . if it does then @xmath365 which shows that @xmath366 . if @xmath367 then by definition @xmath368 with @xmath369 homogeneous of degree zero .
the @xmath370 reduction of @xmath371 is of the form @xmath372\ ] ] for coefficients @xmath373 depending on @xmath374 .
substituting @xmath375 into equation , one has @xmath376 as @xmath363 is homogeneous of degree zero , the exponent of @xmath320 in each term must add to zero , and so @xmath377 since @xmath378 so @xmath379 .
thus one has @xmath380 @xmath381 ( recall the @xmath373 depend on @xmath44 and @xmath203 ) , which shows that @xmath382 as well .
[ t - pr : phi ] there is a unique power series @xmath383 \ ! ] } { [ \ ! [ t ] \ ! ] } $ ] such that @xmath384_{{{h}}}\circ \phi & = \phi \circ [ p]_{{{f}_{0}}}\end{aligned}\ ] ] in @xmath385 \ ! ] } { [ \ ! [ t ] \ ! ] } $ ] .
if @xmath283 in @xmath386 \ ! ] } $ ] is a series with coefficients in a ring @xmath176 of characteristic @xmath6 , let @xmath387 be the corresponding series with coefficients those of @xmath283 , raised to the @xmath364power ; thus @xmath388 in this notation , the equation supposedly satisfied by @xmath389 takes the form @xmath390 where @xmath391 { [ \ ! [ t ] \ ! ] } $ ] is the power series constructed in lemma [ t - le : tau ] . if @xmath392 denotes the compositional inverse ( not reciprocal ) of @xmath360 , and similarly @xmath393 denotes @xmath394 , then this equation can be rewritten in the form @xmath395^{\sigma } \circ \tau^{-\sigma } = \dots.\end{aligned}\ ] ] if @xmath389 is to be of the form @xmath396 then as @xmath397 grows , @xmath398 will converge to the identity in the @xmath399adic topology , and we must have @xmath400 on the one hand , if the limit exists , then it certainly intertwines the @xmath6series . on the other hand ,
the limit exists : lemma [ t - le : tau ] asserts that @xmath401 { [ \ ! [ t ] \ ! ] } , \ ] ] so @xmath402 converges to the identity in the @xmath399adic topology .
the group of formal power series under composition is complete with respect to the non - archimedean norm defined by the degree of the leading term , so the infinite composite converges because the sequence of composita converges to the identity .
it is easy to see in addition that @xmath389 inherits the property @xmath403 from @xmath360 .
[ t - pr : e - phi - is - h ] the power series @xmath389 is the unique strict isomorphism @xmath404 of formal group laws over @xmath385 \ ! ] } $ ] such that @xmath357 let us write @xmath304 for the formal group law @xmath405 where @xmath406 since @xmath15_{{\overline{f } } } ( t ) = t^{q}$ ] , the uniqueness of @xmath407 follows from proposition [ t - pr : phi ] .
we need to show @xmath408 there is a canonical strict isomorphism @xmath409 from a @xmath6typical formal group law @xmath410 , defined over @xmath385 \ ! ] } $ ] .
indeed @xmath411 is given by the formula ( * ? ? ?
* a2.1.23 ) @xmath412_{{\overline{f } } } [ \tfrac{1}{r}]_{{\overline{f } } } \sideset{}{^{{\overline{f}}}}\sum_{i=1}^{r } \zeta^{i}t,\ ] ] where @xmath413 and @xmath414 is a primitive @xmath397th root of @xmath415 .
we first claim that @xmath416 ; then we shall show that @xmath411 is the identity . as @xmath410 and @xmath302
are both are @xmath6typical , it suffices to show that @xmath15_{{g } } = [ p]_{{{h}}}$ ] . since @xmath411 is a homomorphism of groups
, one has @xmath417_{{\overline{f}}}\circ
\rho = \rho\circ [ p]_{{g}}\ ] ] using equation , one has @xmath362_{{\overline{f}}}\circ \rho ( t ) & = \sideset{}{^{{\overline{f}}}}\sum_{p { \nmid}r \geq 1 } [ \tfrac{\mu ( r)}{r}]_{{\overline{f } } } \sideset{}{^{{\overline{f}}}}\sum_{i=1}^{r } [ p]_{{\overline{f } } } ( \zeta^{i}t ) \\ & = \sideset{}{^{{\overline{f}}}}\sum_{p { \nmid}r \geq 1 } [ \tfrac{\mu ( r)}{r}]_{{\overline{f } } } \sideset{}{^{{\overline{f}}}}\sum_{i=1}^{r } \zeta^{iq}t^{q } \\ & = \rho(t^{q})\end{aligned}\ ] ] since @xmath290 and @xmath15_{{\overline{f } } } ( t ) = t^{q}.$ ] thus @xmath418_{{g } } ( t ) ) = \rho(t^{q } ) = \rho([p]_{{{h } } } ( t)).\ ] ] next we must show that @xmath411 is the identity . as @xmath419
we have @xmath420 but @xmath299 is @xmath6typical , so @xmath421_{{{f}_{0 } } } \sideset{}{^{{{f}_{0}}}}\sum_{i=1}^{r } \zeta^{i}t = 0\ ] ] for @xmath422 ; it follows from equation that @xmath423 but equation and @xmath15_{{\overline{f } } } ( t ) = t^{q}$ ] imply that @xmath424 if @xmath425 then we must have @xmath426 together with equation this implies @xmath427 .
for @xmath428 we have @xmath429 \ ! ] } $ ] , so @xmath430 .
if @xmath431 is _ any _ isomorphism from the group law @xmath299 to the group law @xmath302 over @xmath347 , then it must satisfy the equation @xmath119_{{{h } } } \circ \psi = \psi \circ [ p]_{{{{f}_{0}}}}.\ ] ] by lemma [ t - le : tau ] , we must have @xmath432 because @xmath433 the intertwining equation can be rewritten inductively as a sequence of generalized artin schreier equations @xmath434 beginning with @xmath435 because of ( [ define - psis ] ) , the @xmath436 are all algebraic , and the galois group of the extension they generate acts by translating the solutions by an element of the field with @xmath364 elements .
the coefficients of @xmath389 satisfy the same equations in @xmath437 .
starting with @xmath438 , we may adjust each @xmath436 by a galois transformation so that its image in @xmath354 is @xmath439 .
the resulting power series @xmath440 is an isomorphism of formal group laws in @xmath347 , since it becomes one in @xmath354 .
the uniqueness of @xmath440 satisfying the hypotheses is a trivial consequence of the uniqueness of @xmath389 in proposition [ t - pr : phi ] .
it is the field @xmath344 rather than @xmath437 which appears in the tate homology calculations .
expanding a rational function as a power series at zero gives an embedding @xmath441 , which extends to an embedding @xmath442 thus we have [ t - co : phi - at - last ] there is a unique strict isomorphism @xmath443 , satisfying * the coefficients of @xmath353 are in the subfield @xmath347 and * the expansion of @xmath353 at @xmath444 is a power series with coefficients in @xmath385 \ ! ] } $ ] , congruent to the identity modulo @xmath445 .
suppose that @xmath454 is a formal group law of finite height @xmath7 over a field @xmath446 of characteristic @xmath6 , that @xmath447 is a lift of @xmath446 , and that @xmath455 is an @xmath447algebra .
the set of isomorphism classes of deformations to @xmath455 is a functor from @xmath447algebras to sets ; lubin and tate show that this functor is representable .
namely , let @xmath464 \ ! ] } , \\
c_{p^{i } } ( x , y ) & = \frac{1}{p}[(x+y)^{p^{i}}-x^{p^{i}}-y^{p^{i}}],\end{aligned}\ ] ] and let @xmath456 be any deformation of @xmath454 to @xmath25 such that @xmath465 for @xmath466 a unit of @xmath448 and @xmath467 .
lubin and tate show that such deformations exist ; we shall call such a group law a _
lubin - tate lift _ of @xmath454 .
theorem 3.1 of @xcite may be phrased as follows .
[ t - th : lubin - tate ] if @xmath461 is a deformation of @xmath454 to an @xmath447algebra @xmath455 , then there is a unique map of @xmath447algebras @xmath468 such that there is a isomorphism @xmath469 moreover the isomorphism is unique .
a classical theorem of i cohen ( see @xcite ) asserts that for any field @xmath446 of characteristic @xmath6 , there is an essentially unique complete discrete valuation ring @xmath470 with maximal ideal generated by @xmath6 and residue field @xmath446 ; in particular , @xmath470 is noetherian .
if @xmath446 is perfect , then @xmath470 is just the ring @xmath471 of witt vectors of @xmath446 , but in general it is a subring of the witt vectors , and ( although it is a functor ) it is not very easily described .
however , when the degree @xmath472 $ ] is finite ( instead of @xmath415 as in the perfect case ) , eg if @xmath473 , then the cohen ring is still relatively tractable . in this case , for example , we have the @xmath6adic completion of @xmath83 is a complete discrete valuation ring with maximal ideal generated by @xmath6 and residue field @xmath344 , so it is isomorphic to the cohen ring of @xmath344 .
indeed schoeller @xcite shows that an isomorphism is given by a choice of @xmath6base for @xmath344 and representatives of that @xmath6base in @xmath475 .
equation reduces the second part of the lemma to the observation that the rings @xmath476 \ ! ] } \cong \bigl({{\mathbf{z}}_{p } } { [ \ ! [ { { w}}_{1},\dots,{{w}}_{n-2 } ] \ ! ] } { ( \ ! ( { { y } } ) \ ! ) } \bigr){^{\wedge}_{i_{n-1}}}\ ] ] are isomorphic .
it is a standard fact about @xmath478 that @xmath479 for @xmath467 . by the definition of @xmath296 ,
one then has @xmath480 for @xmath481 .
it follows that @xmath296 satisfies equations for the group law @xmath482 .
the formal group law @xmath483 over @xmath484 has image @xmath302 in @xmath303 , so by corollary [ t - co : phi - at - last ] the pair @xmath485 is a deformation of @xmath299 to the @xmath477algebra @xmath486 \ ! ] } .\ ] ] theorem [ t - th : lubin - tate ] provides a ring homomorphism @xmath487 and an isomorphism of formal group laws @xmath488{c } { \mathbf{c}}_{{{{\mathbf{f}_{p } } { ( \ ! ( { { y } } ) \ ! ) } ^{\textup{sep}}}}}\hat{{\underset{{{\mathbf{z}}_{p}}}{\otimes } } } { g_{n-1}}\ ] ] such that @xmath489 . the ring on the left represents deformations of @xmath491 to@xmath84algebras , while the ring on the right represents deformations of@xmath492 to @xmath84algebras .
the isomorphism @xmath389 of corollary [ t - co : phi - at - last ] induces an isomorphism between these functors
. there are isomorphisms @xmath493 \ ! ] } \\ { e_{n-1}}^{0 } ( { \mathbf{c}}p^{\infty } ) & \cong { e_{n-1}}^{0 } { [ \ ! [ t_{2 } ] \ ! ] } \\\end{aligned}\ ] ] such that the formal group law @xmath296 expresses the coproduct on @xmath494 , and the formal group law @xmath483 expresses the coproduct on @xmath495 .
a standard argument @xcite using landweber s exact functor theorem gives : [ t - th : ring - map ] there is a canonical map of ring theories @xmath496 whose value on coefficients is @xmath283 , and whose value on @xmath494 is determined by the equation @xmath497 here @xmath498 is the isomorphism . , * m mahowald * , * h sadofsky * , _ constructions of elements in picard groups _ , from : `` representation theory and algebraic topology , northwestern 1992 '' , m mahowald and e friedlander ( editors ) , volume 158 of contemporary mathematics , ams , providence , ri ( 1994 ) 89126 , _ bousfield localization functors and hopkins chromatic splitting conjecture _ , from : `` the ech centennial , 1993 '' , m cenkl and h miller ( editors ) , volume 181 of contemporary mathematics , ams , providence , ri ( 1995 ) 225250 , _ the elliptic character and the witten genus _ , from : `` algebraic topology ( northwestern university , 1988 ) '' , mark mahowald and stewart priddy ( editors ) , volume 96 of contemporary mathematics , ams , providence , ri ( 1989 ) 281289 , _ the geometric realization of the chromatic resolution _ , from : `` algebraic topology and algebraic @xmath505theory '' , w browder ( editor ) , volume 113 of annals of math .
studies , princeton university press , ( 1987 ) 68 - 179 | we construct splittings of some completions of the @xmath0tate cohomology of @xmath1 and some related spectra . in particular , we split ( a completion of ) @xmath2 as a ( completion of ) a wedge of @xmath3 s as a spectrum , where @xmath4 is shorthand for the fixed points of the @xmath0tate cohomology spectrum ( ie the mahowald inverse limit @xmath5 ) .
we also give a multiplicative splitting of @xmath2 after a suitable base extension . |
gamma - ray lines emitted in solar flares provide information on the composition , spectrum , and angular distribution of accelerated ions that interact in the solar atmosphere @xcite .
they also provide information on the composition and density of the ambient plasma .
@xcite performed the first calculations of the spectral shape of emission lines due to doppler shifts from recoil of excited nuclei after impact by protons and @xmath1 particles .
they found that the @xmath3o line at 6.13 mev has a width of @xmath6 kev ( fwhm ) under bombardment by an isotropic distribution of protons .
this width increases to @xmath7 kev under bombardment by @xmath1 particles .
they also estimated the shift of the line centroid , for a flare near the center of the solar disk , for both a downward isotropic distribution and a downward isotropic distribution limited to angles @xmath8 from the radial direction .
these estimates indicated red shifts of @xmath9 and 40 kev for proton bombardment and @xmath10 and 60 kev for @xmath1-particle bombardment , respectively .
@xcite performed detailed calculations of the shapes and shifts of gamma - ray lines from @xmath2c and @xmath3o that have become the basis of a monte carlo code fundamental to several subsequent publications .
@xcite used updated cross sections in this code to calculate the solar @xmath11be , @xmath11li , and @xmath2c line profiles for isotropic , fan beam , and downward beamed proton and @xmath1-particle distributions having bessel function energy spectra .
they also included contributions to the @xmath2c line from spallation of @xmath3o .
they described how these angular distributions can be distinguished from one another using gamma - ray line measurements made at different heliocentric angles , especially when made with a high - resolution spectrometer .
based on an optical model with parameters determined by accelerator particle data , @xcite developed a computer code that calculated the shapes of the @xmath2c and @xmath3o lines from inelastic collisions of protons .
they showed representative spectra for fan beams , downward beams , and isotropic interacting particle distributions for both thick and thin targets .
@xcite expanded these line - shape calculations by including @xmath1-particle interactions .
@xcite provided the first estimate of the angular distribution of accelerated ions at the sun from measurements of flare - produced neutrons at earth .
they found that the data could only be fit by ions with either an isotropic distribution or a fan beam distribution produced by particles mirroring in a magnetic loop .
@xcite applied these transport calculations to the production of nuclear de - excitation @xmath0-ray lines and calculated the time profiles of decaying phases of high - energy flares under the assumption that the energy release occurred over short time period relative to the emission .
their comparisons with limited flare data indicated that transport effects of the particles in magnetic loops dominated the time profiles .
@xcite applied these transport calculations to determine the shapes of the @xmath11be and @xmath11li lines produced in @xmath1-@xmath12he fusion reactions .
they derived the angular distribution of interacting @xmath1 particles for different amounts of pitch angle scattering caused by mhd turbulence in coronal magnetic loops .
the distribution approaches that of a fan beam when there is negligible pitch angle scattering . with increasing pitch - angle scattering a larger fraction of the @xmath1 particles interact in the downward direction .
because @xcite were only able to compare their calculations with data from a limb flare , they were not able to constrain the amount of pitch - angle scattering encountered by the @xmath1 particles .
@xcite studied the @xmath11be and @xmath11li line profiles in 19 solar flares using the shapes calculated by @xcite .
they found that a downward beam of @xmath1 particles could be ruled out in two disk flares , and that isotropic and broad fan - beam distributions of interacting particles were both consistent with all the data .
recently @xcite have calculated the 4.439 mev @xmath2c line profiles in more detail than has been done previously .
they used new accelerator measurements and have combined the techniques developed by @xcite and @xcite to calculate the profiles over a broad range of interacting proton and @xmath1-particle energies , including contributions to the @xmath2c line from spallation of @xmath3o .
their algorithm enables calculation of the line profiles for different interacting particle spectra , angular distributions , and @xmath1/p abundances . in figure [ fig1 ]
we plot calculated profiles that illustrate the various processes that contribute to the overall shape of the @xmath2c line .
we assume that the proton and @xmath1 particles follow power - law spectra with an index of 4.0 above @xmath5 2 mev / nucleon , are normalized to the same number of interacting particles , and are beamed ( @xmath13 exponential fall - off ) toward the sun .
the calculations have been done for a flare at a heliocentric of 5@xmath14 . in the left panel
we plot the line profiles produced by the inelastic , @xmath2c(_p , p@xmath0_)@xmath2c , and spallation , @xmath3o(_p , x@xmath0_)@xmath2c , reactions .
the inelastic reaction dominates and produces a narrower line profile than the spallation reaction . in the right panel
we plot the inelastic @xmath2c(_@xmath15_)@xmath2c and spallation @xmath3o(_@xmath16_)@xmath2c reactions .
the inelastic @xmath1-particle reactions produce a more complex line profile extending to lower energy than that produced by the protons .
we have some concern about the @xmath1-particle induced line shapes , however , because they are only based on optical model calculations that are most accurate for energies above @xmath520 mev / nucleon .
laboratory measurements are an important priority .
we compare the total contributions to the @xmath2c line profile from proton and @xmath1-particle reactions in figure [ fig2 ] ; the proton and @xmath1-particle distributions are normalized to the same number of incident particles and once again follow a power - law spectrum with index 4 .
the @xmath1-particle induced reactions dominate the line shape for power - law accelerated - particle spectra and @xmath1/proton ratios of @xmath50.5 typically found in flares @xcite . in this paper
we focus on measurements of the angular distributions of flare - accelerated protons and @xmath1 particles that impact ambient @xmath2c , @xmath3o , and @xmath4ne through measurements of de - excitation lines from the recoiling nuclei .
we use spectral data from the _ solar maximum mission ( smm ) _ gamma - ray spectrometer experiment .
although the spectrometer has only moderate spectral resolution it revealed @xmath51% red shifts for flares centered on the solar disk and @xmath53% intrinsic widths for all three lines .
we also compare the @xmath2c line observations with calculations for different assumed angular distributions of interacting and protons and @xmath1 particles .
the nuclear line measurements that we present were derived from 19 solar flares observed by _ smm _ @xcite .
the gamma - ray spectrometer @xcite was exceptionally stable over its full @xmath510 year mission .
it was also well calibrated .
the energy calibration was determined using laboratory sources and was confirmed by observations of unshifted lines : the solar 0.511 mev ( positron annihilation ) and 2.223 mev ( neutron capture ) lines @xcite , and the 4.444 mev ( @xmath17b and @xmath2c ) and 6.13 mev ( @xmath3o ) lines from the earth s atmosphere @xcite .
based on these measurements , we believe that line energies are known to @xmath183 kev .
the instrumental energy - resolution function was also determined using the laboratory calibrations and confirmed by the same flare and atmospheric line observations .
the instrumental resolution was 7.8% ( fwhm ) at 0.511 mev , 4.6% at 2.223 mev , 3.8% at 4.44 mev , and 3.4% at 6.13 mev .
@xcite noted that the measured widths of de - excitation lines in the sum of the 19 flares were broader than predicted by theory for an isotropic distribution of particles .
such broadening could occur if the interacting particle distribution is anisotropic ( thus producing doppler - shifted lines ) because spectra from flares at various heliocentric angles were summed together .
we therefore summed spectra into 5 groups depending on the heliocentric angle of the flare .
the average angle and number of flares in each group are : 5@xmath14 , 1 ( unfortunately , this was a relatively weak flare ) ; 30@xmath14 , 3 ; 43@xmath14 , 5 ; 74@xmath14 , 5 ; 86@xmath14 , 5 .
we plot count spectra in limited energy ranges around the 4.44 mev @xmath2c/@xmath17b , 6.129 mev @xmath3o , and 1.634 mev @xmath4ne de - excitation lines for the five heliocentric angles in figures 3 - 5 .
we fit the count spectra with a power law and a single gaussian - shaped line with energy , width , and amplitude free to vary .
solid lines through the data represent the best fits . for comparison , in figure [ fig6 ]
we also plot the count spectra and fits for the thermal neutron capture line that is formed near rest and should appear at 2.223 mev .
the vertical dotted lines are located at the rest energies of the nuclear lines .
the centroid energies of the de - excitation lines appear to shift to lower energies for flares close to the center of the solar disk .
we summarize the results on line shifts in figure [ fig7 ] where the best fit line energies and uncertainties for the three de - excitation lines and the neutron capture line are plotted as a function of the cosine of heliocentric angle .
the line energies and errors were derived by passing an incident photon spectrum , containing a power law and 22 lines from 0.3 to 8 mev @xcite through the instrument response function , and mapping @xmath19 ; this yielded the same results as we obtained by simply fitting the counts spectrum with a power law and single gaussian over a restricted energy range ( see figures 3 - 5 ) .
the @xmath2c , @xmath3o , and @xmath4ne de - excitation lines appear to be shifted to lower energies by @xmath51% near sun center .
in contrast , the neutron capture line does not exhibit a significant redshift ( @xmath20% ) at sun center .
the higher fitted energies ( @xmath21% ) near the limb may be due to a contribution from other lines , e.g. @xmath22s and @xmath23n that become significant with limb darkening of the 2.223 mev capture line .
the statistical significance of the line data in figures 3 - 5 was also adequate for us to independently fit for the gaussian widths of the @xmath2c , @xmath3o , and @xmath4ne de - excitation lines for some of the heliocentric angles .
we once again fit the count spectra and obtained the line widths and uncertainties using @xmath19 mapping .
the line widths derived from these fits with smoothly varying @xmath19 maps are plotted in figure [ fig8 ] for the three lines .
there is no significant variation in the widths as a function of heliocentric angle .
the weighted means of the line widths ( fwhm ) are 155 @xmath24 19 kev , 167 @xmath24 26 kev , and 40 @xmath24 10 kev for the @xmath2c , @xmath3o , and @xmath4ne lines , respectively
. these are @xmath53% of the line energies .
earlier calculations ( e.g. @xcite ) suggested widths of @xmath52% and 2.6% for the carbon line produced by direct excitation and spallation , respectively , from proton interactions .
the widths are expected to be even broader if accelerated @xmath1 particles play a significant role in line production @xcite .
we have calculated the shape of the @xmath2c de - excitation line for different interacting particle distributions based on the recent work of @xcite .
line shapes were calculated for the following angular distributions : downward beam ( 5@xmath14 exponential fall - off ) ; fan beam following sin@xmath25 where @xmath26 is the angle between the particle velocity and sun center @xcite ; downward isotropic ; and upward isotropic .
these calculations were performed for flares at the heliocentric angles of the five grouped _ smm _ observations . from a comparison of the fluxes in the 1.63 mev ( @xmath4ne ) and 6.13 mev ( @xmath3o ) lines for the flares grouped by heliocentric angle , we obtained power - law spectral indices from @xmath53.9 to 4.5 for accelerated particles having impulsive - flare abundances and an ambient ne / o ratio of 0.25 @xcite . in figure [ fig9 ]
we plot the calculated line profiles for a downward isotropic angular distribution of accelerated particles following a power - law with spectral index 4.0 at the five heliocentric angles .
the solid and dotted curves show the shapes for an assumed accelerated @xmath1/p ratio of 0.5 and 0.1 , respectively .
it is clear that the @xmath1 particles play a dominant role in determining the line shapes when the spectra are this steep .
we note that the line width produced by @xmath1 particles is significantly broader than that produced by protons .
we then folded the calculated line profiles for the particle distributions and a power - law continuum through the _ smm_/grs instrument response function and fit the count data from 3.8 to 4.8 mev .
the goodness of fit , measured using the @xmath19 statistic , is the probability that a fit to a random distribution of numbers about a mean would give a higher value of @xmath19 for the same number of degrees of freedom .
we found that the derived probabilities were not very sensitive to either the index of the power law or the @xmath1/p ratio .
they varied by @xmath2730% for indices ranging from 4 to 5 and @xmath1/p ratios from 0.1 to 0.5 .
we summarize the results of these fits for the flare spectra accumulated at the 5 heliocentric angles in table [ tbl-1 ] for accelerated power - law particle spectra having an index of @xmath54.2 and @xmath1/p ratio of @xmath50.3 ( typical of flare observations ) .
results from the gaussian fits shown in figure [ fig3 ] , where the line width and energy were free to vary , are listed for comparison .
a downward beam of accelerated particles is ruled out by the fits .
both isotropic and broad fan beam distributions also fail to provide acceptable fits to the data .
in contrast , a predominately downward distribution of isotropic interacting particles provides acceptable fits , comparable to that obtained by the gaussian fits . there is a small improvement in some of the fits when an upward isotropic distribution with @xmath520% of the amplitude of the downward isotropic distribution is added ( see probabilities listed in the parentheses ) in column 6 .
the angular distributions that we have used in the fits above are approximations to distributions that are derived under more realistic conditions . @xcite and
@xcite calculated the angular distributions of interacting particles after transport in magnetic loops where the field is constant in the corona and converges from the chromosphere to the photosphere .
this calculation takes into account both magnetic mirroring and pitch angle scattering .
@xcite define a parameter @xmath28 as the scattering mean free path divided by the half length of the coronal segment of the magnetic loop .
they show that pitch angle scattering saturates for @xmath28 = 25 ; particles are not scattered for @xmath29 . in the left panel of figure [ fig10 ]
we plot the angular distributions of selected interacting particle distributions .
they are plotted versus cos @xmath26 , where @xmath26 is the angle relative to the solar radius .
we see that the downward isotropic distribution is a step function in this representation and the fan beam ( sin@xmath25 ) shows a smoothly falling symmetric distribution about the parallel to the solar surface . the weak pitch - angle scattering distribution
, @xmath28 = 300 , begins to approach the fan beam but still has a downward component .
the strong pitch - angle scattering distribution , @xmath28 = 30 , shows a broad downward distribution .
in the right panel of figure [ fig10 ] we plot the calculated @xmath2c line profiles for the four angular distributions plotted in the left panel .
these calculations were done for a flare at a heliocentric angle of 30@xmath14 , an accelerated @xmath1/p ratio of 0.5 , and power law spectral index of 4.0 .
the downward isotropic and @xmath28 = 30 distributions are similar .
their shapes are indistinguishable for the moderate resolution _ smm
_ detectors as reflected in the fits listed in table [ tbl-1 ] .
the weak pitch angle scattering ( @xmath28 = 300 ) distribution has a blue wing that results in a factor of two poorer fit at 30@xmath14 .
the fan beam distribution is significantly different and is clearly incompatible with the _ smm _ line observations .
we have measured the line energies and widths of @xmath0-ray lines produced by excitation of ambient @xmath2c , @xmath3o , and @xmath4ne by solar flare protons and @xmath1 particles .
the @xmath0-ray spectra from nineteen solar flares observed by the _
solar maximum mission _ satellite were grouped by heliocentric angle and were initially fit using gaussian shaped lines .
the goodness of fit was characterized using the @xmath19 statistic and the results are listed in table [ tbl-1 ] .
the gaussians adequately represented the line shapes obtained with the moderate resolution nai detectors .
the energies of each of the lines were consistent with their rest values for flares observed at the solar limb .
the energies of each of the lines decreased by @xmath51% for flares observed near the center of the solar disk .
all three lines exhibit doppler broadening of @xmath53% fwhm of their rest energies ( @xmath2c , 155 @xmath24 19 kev ; @xmath3o , 167 @xmath24 26 kev ; @xmath4ne , 40 @xmath24 10 kev ) .
these results are suggestive of a broad angular distribution of particles primarily interacting in the downward direction .
the line shapes are determined by the location of the flare , the spectra , angular distribution , and composition of the interacting particles , and the intrinsic shapes determined by nuclear physics @xcite .
we therefore fit the same _ smm _ spectral data obtained at the five heliocentric angle with line shapes derived for downward beam , fan beam , downward isotropic , and isotropic interacting particle distributions .
the results of these fits are also given in table [ tbl-1 ] .
a downward beam is clearly ruled out , while a broad fan beam and an isotropic distribution also appear to be inconsistent with the spectral data .
only a downward isotropic distribution gives acceptable fits , comparable to those derived for gaussian shaped lines .
we note that there is some improvement in the fits when we assume a distribution containing downward and upward isotropic components in the ratio of 5:1 .
thus the data are consistent with a very broad angular distribution of particles that primarily interact in the solar atmosphere in the downward direction .
it is important to relate these results on the distribution of interacting particles to their initial acceleration and transport in the solar magnetic loops .
we applied the transport and pitch angle scattering model developed by @xcite and @xcite to calculate angular distributions of interacting particles .
we then folded these distributions into the algorithm that calculates the line shapes @xcite . from table
[ tbl-1 ] we see that a scattering parameter , @xmath28 , of 30 provides comparably good line fits to those obtained for a gaussian and for the downward dominated isotropic distribution .
increasing the scattering parameter ( decreasing the amount of pitch angle scattering ) by a factor of 10 has only a relatively small effect on the goodness of the fits .
it is only when pitch angle scattering completely vanishes , and the particle distribution more closely follows that of a symmetric fan beam parallel to the photosphere , that the model no longer is consistent with the data .
this demonstrates that mhd turbulence is significant in the coronal magnetic loops .
as we have discussed earlier , @xmath1 particles play a significant role in determining the @xmath0-ray line shapes for power - law spectra with indices @xmath304.0 , especially if the accelerated @xmath1/p ratio is @xmath310.1 . in our study of the @xmath2c line shape we noted a small , but not significant , improvement in our fits when we increased the accelerated @xmath1/p ratio to 0.5 .
@xcite presented evidence that the average accelerated @xmath1/p ratio is in fact closer to @xmath50.5 for photospheric @xmath12he abundances in the interaction region @xcite .
@xcite also studied the line shapes of the @xmath1-@xmath12he fusion lines in 19 individual flares and concluded that a downward beam of particles was inconsistent with the data , but that isotropic or fan beam distributions could fit the data .
we have repeated that study using the spectra summed by heliocentric angle .
we find that the spectral resolution of the _ smm _ nai detector is not adequate for us to unambiguously distinguish between isotropic , downward isotropic , and fan beam accelerated particle distributions using the @xmath1-@xmath12he fusion lines .
we expect that the high - resolution spectrometer on board the recently launched _ hessi
_ satellite @xcite will be able to distinguish between these distributions , and will determine if the same angular distribution of accelerated particles is responsible for producing both the @xmath1-fusion and de - excitation lines .
this same spectrometer will also make more definitive measurements of the amount mhd pitch angle scattering in the coronal magnetic loops .
we are grateful to benz kozlovsky and carl werntz for discussions related to calculations of the @xmath2c line shapes .
we also wish to thank xin - min hua for his assistance with calculations of particle transport in magnetic fields and vincent tatischeff for discussions and careful reading of the manuscript .
this work was partially supported by nasa dpr 18995 .
forrest , d. j. , et al .
1980 , solar phys . , 65 , 15 hua , x. -m . & lingenfelter , r. e. 1987 , , 323 , 779 hua , x. -m . , ramaty , r. , & lingenfelter , r. e. 1989 , , 341 , 516 kiener , j. , de srville , n. , & tatischeff , v. 2001 , phys .
c , 64 , 025803 lang , f. l. & werntz , c. w. 1991 , gamma - ray line astrophysics , p. durouchoux & n. prantzos , new york : amer .
physics , 445 lin , r.p .
, et al . 2000 , high energy solar physics - anticipating hessi , r. ramaty & n. mandzhavidze , san francisco : astr .
pacific , 1 - 12 .
mandzhavidze , n. , ramaty , r. , & kozlovsky , b. 1997 , , 489 , l99 murphy , r.j .
1985 , phd dissertation , univ . of md . ,
college park , md .
murphy , r. j. , hua , x. -m . , kozlovsky , b. , & ramaty r. 1990 , , 351 , 299 murphy , r. j. , kozlovsky , b. , & ramaty r. 1988 , , 331 , 1029 ramaty , r. & crannell c. j. 1976 , , 203 , 766 ramaty , r. , kozlovsky , b. , & lingenfelter , r. e. 1979 , , 40 , 487 ramaty , r. , mandzhavidze n. , & kozlovsky , b. 1996 , in : high energy solar physics , r. ramaty , n. mandzhavidze , & x - m .
hua , woodbury , ny : amer .
, 172 share , g. h. & murphy , r. j. 1995 , , 452 , 933 share , g. h. & murphy , r. j. 1997 , , 485 , 409 share , g. h. & murphy , r. j. 2000 , high energy solar physics - anticipating hessi , r. ramaty & n. mandzhavidze , san francisco : astr .
pacific , 377 share , g. h. & murphy , r. j. 2001 , jgr , 106 a1 , 77 werntz , c. , lang , f. l. , & kim , y.e .
1990 , , 73 , 349 rrrrrrrr 5@xmath14&0.35 & 0.12 & 0.19 & 0.22 & 0.20 ( 0.23 ) & 0.21 & 0.22 + 30@xmath14&0.32 & 0.04 & 0.007 & 0.007 & 0.31 ( 0.29 ) & 0.33 & 0.15 + 43@xmath14 & 0.18 & @xmath32 & 0.03 & 0.03 & 0.15 ( 0.24 ) & 0.18 & 0.20 + 74@xmath14 & 0.17 & @xmath33 & 0.22 & 0.22&0.16 ( 0.20 ) & 0.17 & 0.23 + 86@xmath14 & 0.77 & 0.40 & 0.76 & 0.71&0.70 ( 0.71 ) & 0.70 & 0.74 + | the energies and widths of @xmath0-ray lines emitted by ambient nuclei excited by flare - accelerated protons and @xmath1 particles provide information on the ions directionality and spectra , and on the characteristics of the interaction region .
we have measured the energies and widths of strong lines from de - excitations of @xmath2c , @xmath3o , and @xmath4ne in solar flares as a function of heliocentric angle .
the line energies from all three nuclei exhibit @xmath51% redshifts for flares at small heliocentric angles , but are not shifted near the limb .
the lines have widths of @xmath53% fwhm .
we compare the @xmath2c line measurements for flares at five different heliocentric angles with calculations for different interacting - particle distributions .
a downward isotropic distribution ( or one with a small upward component ) provides a good fit to the line measurements .
an angular distribution derived for particles that undergo significant pitch angle scattering by mhd turbulence in coronal magnetic loops provides comparably good fits . |
the transverse ising model ( tim ) is described by a two - state ising hamiltonian with a term representing a field transverse to the spins , @xmath0 where @xmath1 , @xmath2 is the transverse field , @xmath3 and @xmath4 are pauli spin-@xmath5 operators and the first sum is over the nearest neighbors spins on the lattice .
the model has been firstly applied to describe the phase transitions and the properties of hydrogen bonded ferroelectrics @xcite and magnetic ordered materials @xcite .
this model in one dimension has no phase transition at finite temperatures ; however , at zero temperature it is ordered up to the critical value of the transverse field .
the model has been solved exactly in one dimension @xcite . in high dimensions
there are approximations for low - temperatures or high - temperatures regions @xcite .
all other calculations are based on mean field type approximations .
an effective field theory has been presented which improve over mean field results @xcite . since then
many results have been obtained based on the effective field theory .
more recently the model has been used to study the phase diagrams of nanowires systems @xcite and magnetization of nanoparticles @xcite .
the objective of this paper is to present rigorous upper bounds for the critical couplings .
we will apply the results for the d=2 square lattice and the d=3 cubic lattice .
in this section , we describe the methodology used by s barreto et al @xcite to derive an identity for the two - spin correlation function of the transverse ising model .
the procedure used in this deduction was presented in reference @xcite to obtain an exact relation for the order parameter @xmath6@xmath3@xmath7 which generalizes callen s identity @xcite . the longitudinal two - spin correlation function @xmath6@xmath8@xmath7 can be calculated from @xmath9 where h is given by ( [ eq1 ] ) .
the hamiltonian can be separated into two parts , @xmath10 , where @xmath11 includes all parts associated with site i and @xmath12 represents the rest of the hamiltonian .
a direct calculation leads to @xmath13 \delta \big>\label{eq3}\ ] ] where @xmath14 represents the partial trace with respect to site i and @xmath15 .
equation ( [ eq3 ] ) is an exact relation .
however , it is difficult to be used
. therefore , we will make an approximation based on the following decoupling , @xmath16\delta\big > \approx \big < s_l^z\big[\frac{tr_{(i)}s_i^z e^{-\beta h_i}}{tr_{(i)}e^{-\beta h_i}}-s_i^z\big]\big > < \delta>\label{eq4}\ ] ] inserting ( [ eq4 ] ) into ( [ eq3 ] ) and using the fact that @xmath17 , we obtain , @xmath18 by expanding @xmath19 we see that the approximation is correct to the order of @xmath20 .
moreover , it is consistent with the application of the correlation inequalities that will be used later to obtain the upper bounds for the critical couplings . in the next steps we will keep only the = sign of ( [ eq5 ] ) .
let us write @xmath21 , where @xmath22 .
diagonalizing and taking the partial trace over i , we get for the longitudinal spin correlation function , @xmath23 , where @xmath24 , @xmath25 introducing the exponential operator @xmath26 , @xmath27 , we obtain , @xmath28 where f(x ) is given by @xmath29 note that @xmath30 .
expanding the exponential in ( [ eq7 ] ) and considering that @xmath31 , we obtain , @xmath32\big>\nonumber\\ \cdot f(x)\mid _ { x=0}\label{eq9}\ ] ] by a similar procedure
the transverse two - spin correlation function @xmath6@xmath33@xmath7 is obtained , @xmath34\big>\cdot g(x)\mid _ { x=0 } \label{eq10}\ ] ] where g(x ) is @xmath35 the expectation value of @xmath36 is given by , @xmath37\big>\cdot g(x)\mid _ { x=0}\label{eq12}\ ] ]
considering the four neighbours of @xmath38 in ( [ eq9 ] ) , expanding the product , applying the exponential operators appearing in the powers of @xmath39 and @xmath40 in f(x ) , we obtain , @xmath41 where @xmath42 > 0 \nonumber\\ & & b_2 = \frac{1}{8}[f(4 j ) - 2f(2 j ) ] < 0 \nonumber\\ \label{eq14}\end{aligned}\ ] ] and @xmath43 , @xmath44 and @xmath45 are neighbours of @xmath38 and @xmath46 is given by ( [ eq8 ] ) .
after a similar calculation we obtain for the cubic lattice , @xmath47 where , @xmath48 > 0 \nonumber\\ & & b_3 = \frac{1}{2 ^ 5}[f(6 j ) - 3f(2 j ) ] < 0 \nonumber\\ & & c_3 = \frac{1}{2 ^ 5}[f(6 j ) - 4f(4 j ) + 5f(2 j ) ] > 0 \nonumber\\\label{eq16}\end{aligned}\ ] ] and @xmath43 , @xmath44 , @xmath45 , @xmath49 and @xmath50 are neighbours of @xmath38 and @xmath46 is given by ( [ eq8 ] ) .
upper bounds for the critical temperature @xmath51 for ising and multi - component spin systems have been obtained by showing ( for @xmath52@xmath7@xmath51 ) the exponential decay of the two - point function @xcite .
the procedure to obtain these upper bounds for the critical couplings of the tranverse ising model is the following : we start from a two - point correlation function equations ( [ eq13 ] ) and ( [ eq15 ] ) and we make use of griffiths inequalities ( griffiths i , ii ) @xcite and newman and lebowitz inequalities @xcite .
a proof of griffiths inequalities has been given for the xy model with no external field @xcite .
extensions of griffiths - kelly - sherman inequalities to quantal systems , under external fields , both longitudinal and transverse , have been proved @xcite .
the physical reason why the griffiths and similar inequalities are valid for the quantal xy - type hamiltonian is that the off - diagonal interaction , namely , @xmath53 , produce the decrease of the ferromagnetic correlation among the @xmath54-spins , but it is not sufficient big to create a cooperative effect to induce an antiferromagnetic correlation . in other words , one can say that @xmath55 is a dynamical random force acting on z - z correlations @xcite .
we establish the inequality for the two - point function @xmath6@xmath56@xmath7 , @xmath57 which when iterated @xcite implies exponential decay for @xmath58 . from eq.([eq13 ] ) , using griffiths ii ( @xmath6@xmath59@xmath60@xmath61@xmath62@xmath63@xmath7 ) in the second term and considering @xmath64@xmath6@xmath65 , we get , @xmath66 where @xmath67 are neighbours of @xmath38 , and @xmath68 from eq.([eq15 ] ) , using griffiths ii in the second term @xmath69@xmath6@xmath65 , newman s inequality @xmath70@xmath71@xmath72@xmath73@xmath62@xmath74@xmath7 , f are polynomials with positive coefficients ) combined with griffiths i @xmath70@xmath75@xmath76 on the third term @xmath77@xmath7@xmath65 , we get , @xmath78 where @xmath79 are neighbours of @xmath38 , and @xmath80 the two - spin correlation functions appearing in ( [ eq19 ] ) and ( [ eq21 ] ) @xmath81 is the one - dimensional model two - spin correlation function separated by a distance of two lattices sites . for the one - dimensional transverse ising model
the exact value of this function at the critical value @xmath82 is @xcite : @xmath83 where @xmath84 .
for the one - dimensional ising model the exact value of the spin correlation function separated by a distance of two lattices sites is : @xmath85 and @xmath86 .
solid line is the mean field approximation result .
results for our approach evaluated by using the spin correlation function separated by two sites , eq.([eq23 ] ) are given by the dotted ( @xmath87 ) and dashed ( @xmath86 ) lines.,scaledwidth=85.0% ] in the following , we will use result ( [ eq22 ] ) in ( [ eq19 ] ) and ( [ eq21 ] ) to calculate the upper bound for @xmath88 at @xmath89 and result ( [ eq23 ] ) in ( [ eq19 ] ) and ( [ eq21 ] ) to calculate the upper bound for @xmath51 at @xmath90 . evaluating numerically the value of t such that @xmath91 , @xmath92 , we obtain , by sufficient condition ( see eq.([eq17 ] ) ) , the upper bounds for @xmath51 as a function of @xmath2 , shown in figure [ figure1 ] , together with the curve for the mean field results
. we use ( [ eq23 ] ) for the one - dimensional two - spin correlation function in obtaining figure [ figure1 ] .
this curve represents the rigorous upper bounds for @xmath93 .
in particular , the mean field values are @xmath94 and @xmath95 .
the values for @xmath88 and @xmath96 , which are the rigorous upper bounds for @xmath97 and @xmath98 , obtained in the present calculation are : ( a ) d=2 , z=4 ; @xmath99 and @xmath100 , ( b ) d=3 , z=6 ; @xmath101 and @xmath102 . in table
[ tab ] , we compare the results obtained by the effective field calculation ( eft ) @xcite , the high temperature expansion ( hte ) @xcite and the present results for @xmath88 .
.estimatives for @xmath103 for d=2 and d=3 . [ cols="^,^,^",options="header " , ] [ tab ]
in this paper we have obtained rigorous upper bounds for the critical couplings of the transverse ising model .
the procedure was based on an approximation for an exact identity for the two - spin correlation functions and on rigorous inequalities for the spin correlation functions .
the approximated relation for the two - spin correlation function , eq.([eq5 ] ) , used in this procedure , is consistent with the rigorous inequalities , eq.([eq17 ] ) , since both act in the same direction of the inequalities .
the upper bounds were applied for two- and three- dimensional models .
fcsb is grateful for the financial support of capes / brazil which made possible his visit to the ufsj / brasil .
alm acknowledges financial support from cnpq / brazil and fapemig / brazil .
18 de gennes p g , collective motions of hydrogen bonds , 1963 solid st . comm .
* 1 * 132 blinc r and zeks b , dynamics of order - disorder - type ferroelectrics and anti - ferroelectrics,1972 adv . in phys . *
91 * 693 wang y l and cooper b , collective excitations and magnetic ordering in materials with singlet crystal - field ground state,1968 phys . rev . *
172 * 539 pfeuty p , the one - dimensional ising model with a transverse field , 1970 ann .
phys . * 57 * 79 katsura s , statistical mechanics of the anisotropic linear heisenberg model , 1968 phys . rev .
* 1508 suzuki m , equivalence of the two - dimensional ising model to the ground state of the linear xy - model , 1971 phys . lett . * 34 a * 94 elliott r j and wood c , the ising model with a transverse field .
i. high temperature expansion,1971 j. phys .
c * 4 * 2359 pfeuty p and elliott r j , the ising model with a transverse field .
ii . ground state properties,1971 j. phys c * 4 * 2370 s barreto f c , fittipaldi i p and zeks b , new effective field theory for the transverse ising model , 1981 ferroelectrics * 39 * 1103 kaneyoshi t , phase diagrams of a transverse ising nanowire , 2010 j.magn .
magn . mater . *
322 * 3014 kaneyoshi t , magnetizations of a nanoparticle described by the transverse ising model , 2009 j.magn .
mater . * 321 * 3430 callen h b , a note on green functions and the ising model,1963 phys.lett . *
4 * 161 fisher m , critical temperatures of anisotropic ising lattices .
ii . general upper bounds , 1967 phys.rev . * 162 * 480 simon b , correlation inequalities and the decay of correlations in ferromagnets , 1980 commun.math phys . * 77 * 111 brydges d , frolich j and spencer t , the random walk representation of classical spin systems and correlation inequalities , 1982 commun.math.phys . * 83 * 123 gallavotti g , a proof of the griffiths inequalities for the xy model , 1971 studies appl .
math * 1 * 89 newman c , gaussian correlation inequalities for ferromagnets,1975 zeitschriftfur wahrscheinlichkeitstheorie * 33 * 75 suzuki m , correlation inequalities and phase transition in the generalized x - y model , 1973 j. math .
* 12 * 837 .
p and lebowitz j l , correlation inequalities for the quantum spin systems with quenched centered disorder , 2010 j. math . phys . * 51 * 023302 | starting from an exact formal identity for the two - state transverse ising model and using correlation inequalities rigorous upper bounds for the critical temperature and the critical transverse field are obtained which improve effective results . |
this paper has two objectives .
the first objective is to apply algebraic topology to a problem in chemical physics .
the problem , when translated into topology , is a special case of a more general problem which is both natural and elementary to state .
our other objective is an investigation of the general problem , which we hope will be of independent interest to mathematicians . in its barest form ,
the chemical physics problem involves calculating the set of solutions to a system of linear equations parametrized by the circle .
the equations are derived from a finite graph equipped with additional structure , where the graph is a mathematical representation of a molecule .
the set of nontrivial solutions of the system of equations is finite and the problem is to enumerate them , taking multiplicity into account .
the solutions are called _ excitons . _
we refer the reader to the end of this section for some of the historical background and to the chemical physics paper @xcite for a more extensive treatment . to translate to problem into algebraic topology , the equations first need to be converted to a compact form .
the requisite form is a parametrized @xmath1-eigenvalue problem@xmath2 quantum mechanics suggests that for all but finitely many points @xmath3 , equation has no non - trivial solutions , and this will be taken as an axiom .
consequently , the solution to the problem is reduced to that of enumerating with multiplicity the set of points @xmath4 for which has a non - trivial solution .
from the perspective of algebraic topology , an advantage of equation is that it recasts the problem as a kind of intersection theory question in the unitary group @xmath0 : the number of excitons is just the intersection number of the curve @xmath5 with the subspace @xmath6 consisting of unitary matrices having non - trivial ( + 1)-eigenspace .
however , there is a catch : for @xmath7 , the space @xmath8 is not a manifold .
nevertheless , @xmath8 is a stratified space and the intersection problem turns out to be tractable .
the intersection problem alluded to above is a special case of a more general problem , whose investigation is the other goal of this paper .
suppose @xmath9 is a closed smooth oriented manifold of dimension @xmath10 equipped with a filtration @xmath11 giving an oriented stratifold structure @xcite .
in particular , the open strata @xmath12 are smooth manifolds of dimension @xmath13 . concretely , one encounters such a filtration if @xmath9 is equipped with the structure of a cw complex , the filtration is given by the skeleta and the characteristic maps for the cells satisfy a smoothness condition .
examples of this kind arise in morse theory @xcite , @xcite . for a fixed degree @xmath14 ,
set @xmath15 . on occasion ,
the singular homology group @xmath16 is found to be infinite cyclic .
is connected and the boundary homomorphism @xmath17 has non - trivial kernel . ] when this happens , the stratifold orientation determines a preferred generator .
denote the generator by @xmath18 .
suppose that @xmath19 is a smooth map where @xmath20 is a closed oriented manifold of dimension @xmath21 .
let @xmath22\in h_p(p;\bbb z)$ ] denote the fundamental class .
then the _ homological intersection number _ ( or _ global intersection index _ ) is defined as @xmath23 ) \in \bbb z \ , , \ ] ] where @xmath24 is the intersection pairing .
henceforth , assume that the preimage @xmath25 is a finite set of points .
is there a description of the homological intersection number given in terms of the restriction of @xmath26 to an arbitrarily small neighborhood of @xmath27 ? stated another way , the question asks for a formula for the global invariant in terms of local invariants coming from the differential topology of the map @xmath26 near the set @xmath27 . in certain cases , the answer to
the above question is well - known and classical .
for example , if @xmath28 happens to be a closed submanifold with fundamental class @xmath18 and @xmath26 happens to be transverse to @xmath29 , then the homological intersection number is just the sum over the points of @xmath27 counted with sign .
however , in the case addressed in this paper , @xmath29 is not a manifold and the function @xmath26 is not assumed to satisfy any transversality . in the case considered in the paper , @xmath9 is @xmath0 , the unitary group of @xmath30 matrices , a manifold of dimension @xmath31 .
we equip @xmath0 with a cw decomposition whose @xmath32-skeleton is the space @xmath33 described above .
then @xmath34 ( for @xmath35 ) is infinite cyclic with preferred generator @xmath36 .
if @xmath7 , then @xmath29 is not a manifold ; see example [ ex : u2 ] below .
a main result ( theorem [ bigthm : indexthm ] ) of this paper is a kind of _ index theorem_. the crucial difference with the classical case is that the sign of a point @xmath37 is now replaced by a _
local index _
@xmath38 , given by the net change in the number of eigenvalues of @xmath39 passing through @xmath40 as nearby points @xmath41 pass through @xmath42 in a counterclockwise direction ( see [ subsec : local ] for details ) .
[ bigthm : indexthm ] assume @xmath43 is smooth and the preimage @xmath44 is discrete .
then the global intersection index of @xmath26 equals the sum of the local indices , i.e. , @xmath45 in summary , theorem [ bigthm : indexthm ] provides a description of the homological intersection number as a sum over local indices in the particular case of @xmath46 . the extent to which a result of this kind applies in other instances remains open .
the molecules we investigate possess discrete , 1-dimensional translational symmetry which is broken only in a finite number of places .
the molecules are mathematically represented by simple metric graphs whose edge lengths are positive integers ( cf . figure [ side - by - side ] ) .
let @xmath47 be a connected , finite , simple graph .
let @xmath48 denote the set of vertices and @xmath49 the set of edges . the graph @xmath47 models the geometry of our molecule .
since @xmath47 is simple , there is at most one edge for each pair of distinct vertices .
if @xmath50 are distinct vertices joined by an edge , we often refer to the edge by @xmath51 . associated with @xmath47
is a directed graph @xmath52 called the _
double _ of @xmath47 , whose set of vertices is @xmath48 and where a directed edge is given by a pair @xmath53 such that @xmath54 is incident to @xmath55 .
each directed edge is uniquely represented by an ordered pair of distinct vertices @xmath56 and we often specify @xmath55 as @xmath57 ( the arrows of figure [ side - by - side ] depict the directed edges of @xmath52 ) .
the lengths of the edges are given by a function @xmath58 by slight abuse of notation , consider this as a function defined on @xmath59 defined by @xmath60 , with @xmath61 . from the viewpoint of chemical physics ,
the value @xmath62 is the number of repeat units of the edge @xmath57 and corresponds to the number of periods of the discrete symmetry on the edge .
if @xmath63 is a vertex , let @xmath64 denote the set of incoming edges at @xmath54 , and similarly let @xmath65 be the set of outgoing edges at @xmath54 .
there are canonical bijections @xmath66 , where @xmath67 is the set of edges of @xmath49 that meet @xmath54 .
the bijection @xmath68 is given by by @xmath69 and the bijection @xmath70 is given by @xmath71 .
for a finite set @xmath72 , let @xmath73 $ ] be the complex vector space with basis @xmath72 .
let @xmath74 denote the unitary group of @xmath73 $ ] consisting of @xmath75-linear transformations @xmath76\to
\mathbb c[t]$ ] , such such that @xmath77 , where @xmath78 is the conjugate transpose of @xmath79 .
the vertices of @xmath52 break the discrete symmetry , and hence , cause the excitons to scatter ( vertices are often referred to simply as scattering centers for this reason ) .
associated to each vertex @xmath80 , and each real number @xmath81 , there is a _ scattering matrix _
@xmath82 \to \mathbb c[x_1^{a,+}]\,\ ] ] that describes the outgoing waves in terms of the incoming ones .
is known as _ quasimomentum _ ; it appears as a consequence of translational invariance . ]
the scattering matrix is invertible and the function @xmath83 is @xmath84-periodic . in view of the canonical bijections @xmath66 , the scattering matrix at @xmath85
can equally be regarded an invertible self - map @xmath86 \to \mathbb c [ y_1^{a}]\ , .\ ] ] [ hypo : analytic ] for each @xmath87 and @xmath85 , the operator of is unitary .
furthermore , the map @xmath88 given by @xmath89 is real analytic . in particular , the map @xmath90
\to \bbb c [ y_1^{a}]\ , , \ ] ] given by @xmath91 , is real analytic . the hypothesis , which we take as an axiom , is suggested by quantum mechanics @xcite .
it has been substantiated in a variety of cases .
periodicity implies that @xmath92 factors as a map through the circle .
i.e. , @xmath93 where @xmath94 is also real analytic .
excitons satisfy _ time - reversal symmetry ( kramers symmetry ) _ :
@xmath95 in terms of @xmath96 , this equation translates to @xmath97 , which means that @xmath96 is @xmath98-equivariant with respect to conjugation on the circle and inversion on @xmath99 .
the scattering matrices are computed using quantum chemistry methods , and we treat these as known . for a vector
@xmath100 $ ] and a directed edge @xmath101 , let @xmath102 be the component of @xmath103 along @xmath57 .
hence , @xmath104 .
exciton scattering theory implies the following two equations , which are taken here as axioms ( * ? ? ?
( 3),(4 ) ) : @xmath105 where in the indexing for the sum the vertex @xmath54 is fixed .
there is such a pair of equations for every edge in @xmath52 .
the entire system is referred to as the _ es equations .
_ the matrix entries @xmath106 can be interpreted as the amplitude of the wave function @xmath103 on edge @xmath107 as the result of scattering a gaussian plane wave from edge @xmath108 onto vertex @xmath54 @xcite .
the first equation in ( [ eq : esold ] ) expresses the fact that when an exciton wave propagates along a linear segment , it acquires a phase change .
the phase change depends on both the length of the segment and @xmath109 .
the second equation in ( [ eq : esold ] ) connects the amplitude of outgoing waves to those of the incoming . the es equations of ( * ? ? ?
( 3),(4 ) ) are written in a slightly different notation than the equations of .
the notation in the former case is intended for an audience of chemists and physicists , whereas is geared toward mathematicians . to bring algebraic topology into the picture , the equations need to be put into a compact form .
let @xmath110 be the composition @xmath111 where @xmath112 is the diagonal and the map @xmath113 is defined by block sum of matrices ( here , we used the canonical bijection @xmath114 ) .
choose once and for all a linear ordering on @xmath48 .
then the left - lexicographical ordering on ordered pairs identifies @xmath59 with the standard ordered basis for @xmath115 where @xmath116 . in this way
, @xmath110 becomes a @xmath84-periodic real analytic map @xmath117 .
passing to the circle , one obtains a map @xmath118 defined by @xmath119 let @xmath120 { \rightarrow}\mathbb c[x_1]\ ] ] be the length rescaling operator given by @xmath121 .
finally , define a map @xmath122 by @xmath123 one verifies by direct calculation that the full set of exciton scattering equations is equivalent to the single parametrized eigenvector equation @xmath124 where @xmath125 is real analytic , and @xmath126 is a vector . henceforth , we resort to this version of the es equations throughout the paper .
the analyticity of @xmath127 implies that equation can hold only for finitely many points @xmath4 .
specifically , let @xmath128 be the unit sphere . by hypothesis [ hypo : analytic ] ,
the map @xmath129 given by @xmath130 is analytic . by the principle of permanence and the compactness of @xmath131 , the set of zeros
@xmath132 is finite .
the image of the first factor projection @xmath133 is then a finite set of points @xmath134 this is just the set of points @xmath41 where @xmath135 has a non - trivial solution .
let @xmath136 . for @xmath137 ,
let @xmath138 denote the dimension of the @xmath1-eigenspace of @xmath139 .
then @xmath138 is the number of linearly independent solutions of the es equations at @xmath140 .
recall the problem is to count the total number of excitons weighted with their multiplicities , i.e. , @xmath141 .
the following result , which is an application theorem [ bigthm : indexthm ] , provides a lower bound for @xmath10 .
[ bigthm : lower - bound ] let @xmath10 denote the total number of solutions of the es equations , weighted with their multiplicities .
then @xmath142 where @xmath96 is the scattering matrix at the vertex @xmath54 and @xmath143 is the degree of the map @xmath144 .
there is evidence that the inequality of theorem [ bigthm : lower - bound ] is sharp .
in fact , for molecules with large segment lengths , one has the following result .
[ bigthm : long - arm ] up to second order approximation and for sufficiently large segment lengths , the bound of theorem [ bigthm : lower - bound ] becomes an equality .
`` up to second order approximation '' refers to exchanging the one - parameter family of operators @xmath145 with its associated 1-jet at the solution set .
empirical evidence from physics indicates that the addendum is valid even without resorting to the approximation .
the theory of excitons was launched by frenkel , peierls , and wannier in the 1930s @xcite . in its early years
the subject grew slowly and sporadically .
the theory blossomed in the 1950s as a result of an increase in technology as well as the drive for a better understanding of the optical properties of solids @xcite . for an early survey ,
see the introduction to @xcite .
the sort of excitations considered here are produced when an organic structure , like chlorophyll , absorbs light of a certain wavelength .
the absorbed photon excites an electron into a higher energy state and the excited electron can form a bound state with the `` hole '' it leaves behind .
the pair consisting of the excited electron and the hole is known as an _
exciton_. finding a method to accurately enumerate excitons in an organic molecule has consequences for engineering .
for example , it plays an important role in the design of more efficient organic solar cells .
the exciton scattering ( es ) approach , introduced in @xcite ( see also @xcite ) , was a dramatic improvement over previous attempts to study electronic excitations in branched conjugated molecules for a number of reasons ( for an extensive bibliography , see @xcite ) .
the es approach has shown itself to be more powerful than traditional quantum chemistry methods , such as time dependent - density functional theory ( tddft ) : the computational time of electronic excitations with the es approach is on the order of seconds , whereas the same computations with traditional methods are on the order of hours or days . in earlier papers ,
the es approach was only applied to molecules with perfect symmetries , e.g. an ` x ' or ` y ' joint with the same number of repeat units on each arm @xcite .
for example , the molecule displayed in figure [ side - by - side ] fails to have perfect symmetry and is therefore not capable of study using previous techniques .
the symmetric case was done by studying the equations using the full symmetry group of the molecule ( and the representation theory thereof ) .
rewriting the es equations in the form above allows for the analysis of electronic excitations of general branched , conjugated molecules done in this paper , as well as in @xcite , regardless of symmetry
. stated differently , this paper is the result of analyzing the es equations for generic molecules , in a mathematically rigorous way .
theorem [ bigthm : lower - bound ] exhibits an estimate for @xmath10 , the number of solutions to the es equations weighted by their multiplicities .
chemical physicists are interested instead in a number , denoted by @xmath146 , which is the number of excitations that lie in a fixed energy range , called the _
exciton band_. the two numbers are related by the formula @xmath147 where @xmath148 and @xmath149 is the dimension of the @xmath150-eigenspace of @xmath151 for @xmath152 .
( 12 ) ) ) .
the formula is a reflection of the observation that the solutions of the es equations with @xmath153 come in pairs , by time - reversal symmetry .
the appearance of the terms @xmath154 account for the identity @xmath155 , when @xmath156 , which is again a consequence of time - reversal symmetry .
the fundamental domain @xmath157 $ ] is known as the _ brillouin zone .
_ we have written the paper so that it is accessible to readers with a modest background in algebraic and differential topology .
the material from chemical physics especially the technical jargon has been kept to a bare minimum and is usually relegated to remarks or footnotes in an effort to make the paper accessible to a broad mathematical audience .
the cw decomposition of @xmath0 is described in [ subsec : steenrod ] .
the technical results required for the proof of theorem [ bigthm : indexthm ] appear in [ sec : indexthm ] .
an exposition of the proof theorem [ bigthm : indexthm ] appears in [ sec : proof - index - thm ] .
the proof of theorem [ bigthm : lower - bound ] appears in [ sec : lower - bound ] and the proof of addendum [ bigthm : long - arm ] appears in [ sec : long - arm ] . the authors wish to thank tian shi for helpful discussions and for assistance with the figures .
the authors also wish to thank a diligent referee whose numerous suggestions improved the readability of this document .
this material is based upon work supported by the national science foundation grant che-1111350 and the simons foundation collaboration grant 317496 .
we describe the cw decomposition of @xmath0 appearing in steenrod s book @xcite . the decomposition is attributed to miller @xcite and yokota @xcite
. following ( * ? ? ?
iv ) , let @xmath158 be the subspace consisting of vectors whose final coordinates are real and nonnegative , where we are using the standard ordered basis on @xmath159 .
the circle @xmath160 acts freely on @xmath161 by @xmath162 .
let @xmath163 be the quotient space of @xmath131 where @xmath164 for @xmath165 , and @xmath166 for all @xmath167 .
the space @xmath168 is homeomorphic to the reduced suspension @xmath169 , where @xmath170 is the projective space of @xmath115 and @xmath171 denotes the effect of adding a disjoint basepoint to a space @xmath52 .
a point of @xmath168 will be written as an equivalence class @xmath172 $ ] , where @xmath173 and @xmath174 . for each @xmath175
, there is a natural map @xmath176 { \rightarrow}s^{2j-1 } \times s^1 { \rightarrow}q_j \ , , \ ] ] where the second displayed map is the projection onto orbits , and the first is the product of the inclusion @xmath177 with the map @xmath178 { \rightarrow}s^1 $ ] given by @xmath179 . the displayed map defines a relative homeomorphism @xmath180 the maps define a cw decomposition of @xmath168 which is just the suspension of the usual cw structure of @xmath181 .
an embedding of @xmath168 into @xmath0 is given by sending an equivalence class @xmath172 \in q_n$ ] to the transformation which fixes the orthogonal compliment of @xmath182 and multiplies any point on the line @xmath183 by @xmath184 ( by convention , @xmath185 maps to the identity transformation ) .
this gives rise to embeddings @xmath186 for @xmath187 using the inclusion @xmath188 , given by including the first @xmath13 coordinates .
multiplying the cells leads to _ normal cells _ : for any ordered subset @xmath189 there is a map @xmath190 where the first map of is the product of the maps and the second map is given by multiplication .
the normal cells ( together with the 0-cell corresponding to the identity ) give a cw decomposition of @xmath0 ( * ? ? ?
iv , thm .
2.1 ) . by construction ,
the interior of normal cells are identified with unitary transformations characterized by a fixed number of eigenvalues . identifying @xmath191 with its image under the embedding
@xmath186 implies that @xmath191 consists of matrices with @xmath192 along the diagonal , except in the @xmath193 entry . more generally , taking products of such yields @xmath194 , corresponding to matrices with @xmath192 along the diagonal , except in the @xmath195 entries .
the above equips @xmath0 with a cw decomposition whose @xmath32-skeleton is given by the transformations having nontrivial @xmath1-eigenspace , i.e. , @xmath8 .
in particular , the cell @xmath196 is the unique @xmath32-cell . [
ex : u2 ] let @xmath197 .
the above equips @xmath198 with a cw structure with a single cell in dimensions @xmath199 and @xmath200 .
the @xmath201-skeleton in this case is given by @xmath202 , where @xmath27 denotes the reduced suspension .
the space @xmath203 can be thought of as the quotient space of @xmath204 obtained by identifying a single pair of antipodal points , as shown in figure [ fig : rendering ] below .
there is another cw structure on @xmath198 which uses the well - known homeomorphism @xmath205 .
this cw structure also has a single cell in dimensions @xmath199 and @xmath200 , and the @xmath201-skeleton in this case is @xmath206 .
although @xmath203 is homotopy equivalent to @xmath207 , the two spaces are not homeomorphic .
hence the two cw structures are distinct ( in fact , @xmath208 fails to be a subcomplex of the first cw structure ) .
blurring the distinction between the two cw structures in this paper invariably leads to unrecoverable errors .
in particular , the proof of @xmath209 case of proposition [ prop : mult_in_hom ] goes wildly wrong were we to use @xmath207 in place of @xmath210 .
in a previous version of this manuscript , the above cw decomposition was derived from morse theory .
the standard height function @xmath211 on @xmath0 is a morse function for almost all @xmath212 . in a private communication ,
liviu nicolaescu pointed out that for @xmath213 diagonal with distinct , real entries , the flow generated by @xmath26 satisfies the morse - smale condition and leads to the above cw decomposition @xcite .
he also pointed out that in the lagrangian interpretation of the intersections we are considering , @xmath8 can be interpreted as the maslov variety and the global intersection index @xmath214 ( defined below ) is known as the _ maslov index_. however , we will not pursue this interpretation here .
the cw decomposition of @xmath0 given in
[ subsec : steenrod ] has one top cell and its @xmath32-skeleton is @xmath8 .
the top cell of @xmath8 defines a generator @xmath215 ( with integer coefficients being understood ) .
alternatively , the generator can be obtained by means of poincar duality : the fundamental class @xmath216 \in h_{n^2}(u(n))$ ] defined by the top cell of @xmath0 gives an isomorphism @xmath217 < \cong < h^1(u(n ) ) \cong { \mathbb{z}}\ , , \ ] ] where the first displayed map is induced by the inclusion @xmath218 and the second displayed map is poincar duality .
the generator of @xmath219 comes from the determinant map @xmath220 .
the map @xmath221 is an isomorphism by the homology long exact sequence of the pair @xmath222 .
let @xmath223 \in h_1(s^1)$ ] be the fundamental class .
the _ global intersection index _ of a map @xmath46 is the homological intersection number @xmath224 ) \in h_0(u(n);\bbb z ) \cong \bbb z.\ ] ] the _ intersection set _ is the fiber product @xmath225 the set @xmath226 is discrete , i.e. , @xmath227 is a finite set
. it will be convenient to express the global intersection index in terms of orientation classes @xcite . for a subset @xmath228 ,
let @xmath229 note the case @xmath230 is the pair @xmath231 .
the case @xmath232 is a point gives rise to the _ local ( singular ) homology _ at @xmath233 , i.e. , @xmath234 .
let @xmath52 be a smooth @xmath10-manifold and let @xmath235 be the diagonal .
an _ orientation _ for @xmath52 is a ( singular cohomology ) class @xmath236 such that for all @xmath237 , the class @xmath238 generates @xmath239 , where @xmath240 is the inclusion @xmath241 the terminology is not universal .
an orientation in the above sense is equivalent to choosing a thom class for the tangent bundle @xmath242 . in @xcite , @xmath243 is called a _ fundamental cohomology class_. a choice of orientation class @xmath244 is equivalent to a choice of fundamental class for @xmath245 .
henceforth , @xmath243 is chosen to correspond to the class @xmath216 $ ] defined above .
let @xmath246 be the evident map .
[ lem : globalint ] the global intersection index coincides with the kronecker index , i.e. , @xmath247)\rangle$ ] .
this follows immediately from the standard description of the intersection number as a slant product ( cf .
@xcite , @xcite ) .
the local multiplicity , defined below , counts the dimension of the solution space of the equation @xmath248 at a given @xmath41 .
[ defn : multiplicity ] let @xmath249 be any point .
the _ local multiplicity _ of @xmath233 , denoted @xmath250 , is the dimension of the ( + 1)-eigenspace of @xmath251 .
in particular , @xmath252 . the _ local multiplicity _ of @xmath26
is the number @xmath253 where the sum is indexed over all points @xmath254 .
the local multiplicity of @xmath26 enumerates the entire set of solutions to equation , where each solution @xmath233 is counted with multiplicity @xmath250 . for @xmath255 ,
let @xmath256 be the unique point such that @xmath257 . by hypothesis ,
the non - trivial solutions to the equation @xmath258 occur at isolated points @xmath4 .
choose a small arc @xmath259 containing @xmath260 such that the eigenvalues of @xmath39 other than @xmath192 lie outside @xmath259 .
consider an interval @xmath261 centered at @xmath262 .
if @xmath263 , then the number of eigenvalues of @xmath264 which lie in @xmath259 is constant in @xmath265 provided that @xmath266 is chosen sufficiently small . if @xmath267 then the number of eigenvalues of @xmath264 lying in @xmath259 and having positive imaginary part is a constant function on each component .
[ defn : iota - pm ] with respect to the above assumptions , let @xmath268 be the number of eigenvalues lying in @xmath259 having positive imaginary part taken at @xmath269 with @xmath263 .
similarly , let @xmath270 be the number of eigenvalues lying in @xmath259 having positive imaginary part taken at some @xmath271 with @xmath263 .
the _ local intersection index _ at @xmath272 is the integer @xmath273 the _ local intersection index _ of @xmath46 is the sum @xmath274 intuitively , @xmath275 is the number of eigenvalues passing through @xmath276 counterclockwise , as @xmath265 passes through @xmath262 from left to right .
the definition trivially implies the inequalities @xmath277 ( cf . definition [ defn : multiplicity ] ) .
the following result plays a key role of the proof of theorem [ bigthm : indexthm ] . [ prop : mult_in_hom ] there is a preferred isomorphism @xmath278 in particular , @xmath279 is free abelian of rank @xmath250 , so the local multiplicity of @xmath272 is detected by local homology .
assume that @xmath7 , as the @xmath280 case is trivial .
let @xmath281 be a point having multiplicity @xmath250 , i.e. , the dimension of the @xmath1-eigenspace of @xmath282 is @xmath250 . by the knneth theorem of pairs
, it will be enough to exhibit a preferred isomorphism @xmath283 ( cf .
[ fig : rendering ] ) . fix @xmath284 and take a standard neighborhood @xmath285 using the standard operator norm on the lie algebra @xmath286 ( i.e. @xmath287 matrices @xmath79 satisfying @xmath288 ) .
let @xmath289 where @xmath290 is the exponential map .
this produces a neighborhood of any @xmath291 .
define @xmath292 , to obtain a neighborhood of @xmath293 .
the closure of @xmath294 is contained in the interior of @xmath295 , so excision applies in this situation and produces an isomorphism @xmath296 by definition , @xmath297 is a sphere of radius @xmath298 in @xmath0 centered at @xmath282 ( since @xmath299 is a local diffeomorphism ) , so @xmath300 .
apply alexander duality to obtain a preferred isomorphism @xmath301 where @xmath302 denotes reduced cohomology . in , we are using that @xmath303 is homotopy equivalent to @xmath304 , and that @xmath304 is contractible for every @xmath293 to @xmath305 , and @xmath259 to @xmath306 , respectively . ] .
the unreduced cohomology group @xmath307 is free abelian on the number of components of @xmath308 .
let @xmath309 denote the set of these components .
we will first show that @xmath310 . the space @xmath311 consists of transformations with no eigenvalues equal to one , and is open in @xmath312 ( since @xmath313 is closed ) .
recall the arc @xmath259 associated to @xmath233 defined at the beginning of this subsection .
each point @xmath314 has precisely @xmath250 eigenvalues in the arc @xmath259 , none of which are equal to 1 .
let @xmath315 be the function which assigns to each @xmath316 the number of eigenvalues lying in the arc @xmath259 with positive imaginary part .
then @xmath317 is connected and non - empty since it contains @xmath318 , where @xmath319 is a diagonal , skew - symmetric matrix whose first @xmath13 entries are @xmath320
. then @xmath321 ( see figure [ fig : nbhd_u ] ) . hence , @xmath322 has @xmath323 components .
the subspaces @xmath324 determine a set of preferred generators @xmath325 for @xmath326 .
we infer that the classes @xmath327 form a basis for the reduced cohomology group @xmath328 .
let @xmath329 with @xmath330 be the homology class corresponding to @xmath331 via alexander duality .
then @xmath332 is a basis for @xmath333 .
this basis gives the identification @xmath334 .
( 2,2 ) circle ( 2 cm ) ; ( 3,2 ) circle ( 1 cm ) ; at ( 4,2 ) [ circle , draw = black , fill = black , inner sep=0.5 mm , label = right:@xmath233 ] ; ( 1,2 ) ellipse ( 1 cm and 0.5 cm ) ; ( 3.28,3.2 ) ellipse ( 0.25 cm and 0.13 cm ) ; ( 3.28,0.8 ) ellipse ( 0.25 cm and 0.13 cm ) ; [ rem : add - to - prop ] with respect to the isomorphism of proposition [ prop : mult_in_hom ] and the identification @xmath336 , it is straightforward to check that the evident homomorphism @xmath337 corresponds to the diagonal homomorphism @xmath338 .
suppose that @xmath339 is a map of pairs such that @xmath340 is a closed subspace and @xmath341 , where @xmath342 is disjoint from @xmath79 .
suppose also that @xmath52 is a normal space .
then there is an open neighborhood @xmath343 of @xmath79 which is disjoint from an open neighborhood of @xmath342 .
it follows that @xmath344 determines a map of pairs @xmath345 . by excision ,
hence there is an induced homomorphism @xmath347 let @xmath348 .
apply the construction to the map of pairs @xmath349 to obtain a homomorphism @xmath350 consider the composition @xmath351 & h_{n^2}(d_1u(n ) \times s^1\ , |\ , p ) \ar[d]^{(j\times f|p ) _ * } \\ & h_{n^2}(u(n ) \times u(n ) \,|\delta ) \ar[r]^(.75){\langle x,{-}\rangle}_(.75)\cong & \bbb z\ , , } \ ] ] where @xmath352 is the orientation class .
the unlabeled arrow is induced by the evident map @xmath353 .
[ prop : losum ] the homomorphism maps the class @xmath354 $ ] to the local intersection index @xmath275 . the unlabeled homomorphism appearing in the displayed composition is identified with the diagonal @xmath338 ( cf .
remark [ rem : add - to - prop ] ) .
we need to identify the homomorphism @xmath355 of , which is of the form @xmath356 .
if @xmath357 generates the @xmath358-th factor of @xmath359 , @xmath360 , then the value of this homomorphism on @xmath357 is an integer @xmath361 .
it suffices to show @xmath362 .
an artifact of the argument given below is that @xmath363 for at most two of the indices .
recall the neighborhoods @xmath364 and @xmath365 from the proof of proposition [ prop : mult_in_hom ] . to avoid notational clutter , eliminate the argument @xmath282 from the notation and write @xmath366 and @xmath367 respectively . for @xmath368 with @xmath369 , define @xmath370 to be an open arc of length @xmath371 centered at @xmath3 , where @xmath372 is chosen sufficiently small so that @xmath373 .
consider the homomorphism @xmath374 clearly , it will be enough to identify since it is a localized form of . if @xmath375 is the inclusion , then the pushforward of the fundamental class of the circle , i.e. , @xmath376 ) \in h_1(w\ , |\ , z ) \cong \bbb z\ , , \ ] ] is a generator . by the proof of proposition [ prop : mult_in_hom ]
, the domain of has a preferred isomorphism to @xmath359 , where @xmath250 is the multiplicity of @xmath264 .
following that proof , there are generators @xmath377 for the domain . applying the orientation class to the image of these generators
yields the intersection product @xmath378 . to analyze this product ,
we make use of the fact that the homological intersection product is equivalent to a linking number ( * ? ? ? * , ex . 2 , 3 ) .
therefore , @xmath378 is non - zero only when the curve @xmath379 links with @xmath380 , which occurs when @xmath379 has non - trivial intersection with the path component of @xmath381 corresponding to the class @xmath382 via alexander duality .
in general , this intersection index will only be non - zero on at most two of the generators ( see figure [ fig : nbhd_u ] ) .
let @xmath79 be the arc @xmath383 .
the orientation of the arc @xmath384 is induced by the counter - clockwise orientation of @xmath385 .
the counterclockwise orientation of @xmath385 implies that the intersection index will be given by the number of eigenvalues of @xmath386 in @xmath79 minus the number of eigenvalues of @xmath387 in @xmath79 .
but this is the definition of the local intersection index @xmath275 .
( 2,2 ) circle ( 3 cm ) ; ( 2.52,4.95 ) .. controls ( 2,3.1 ) .. ( 2,2 ) .. controls ( 1.3,2.5 ) .. ( .97,4.82 ) arc ( 110:80:3 cm ) ; ( .732,-.72 ) .. controls ( 1,-.5 ) .. ( 2,2 ) .. controls ( 2.5,1.1 ) .. ( 2.78,-.898 ) arc ( 285:245:3 cm ) ; ( 5,2 ) .. controls ( 3,2.5 ) .. ( 2,2 ) ; ( 4.12 , 4.12 ) .. controls ( 3,3 ) .. ( -0.12 , -0.12 ) ; ( -.6,3.5 ) .. controls ( 0,2.5 ) .. ( 2,2 ) ; ( -.75,.78 ) .. controls ( 0,2 ) .. ( 2,2 ) ; ( -1,2 ) .. controls ( -.5 , 2.3 ) .. ( 2,2 ) ; ( 2,2 ) .. controls ( 4,1 ) .. ( 3.93 , -.3 ) ; at ( -1,2 ) [ circle , draw = red , fill = red , inner sep=1 mm ] ; at ( 3.93,-.3 ) [ circle , draw = red , fill = red , inner sep=1 mm ] ; at ( 2,2 ) [ circle , draw = black , fill = black , inner sep=0.7 mm ] ; at ( 5,0 ) @xmath388 ; at ( 5.5,3 ) @xmath389 ; at ( 3.5,5 ) @xmath390 ; at ( 0,4.9 ) @xmath391 ; at ( -1.8,2 ) @xmath392 ; at ( -1.3,0.4 ) ;
the proof of theorem [ bigthm : indexthm ] relies on the commutative diagram @xmath393_{a_\ast } \ar[r]^{(j \times f)_\ast } \ar[d ] & h_{n^2}(u(n ) \times u(n ) ) \ar[d]^{b_\ast } \\
h_{n^2}(d_1 u(n ) \times s^1\ , |\ , \mathfrak i_{f } ) \ar[d]_\cong \ar[r]^{(j \times f|\mathfrak i_f)_\ast } & h_{n^2}(u(n ) \times u(n)\ , |\ , \delta ) \\ \bigoplus_{p \in \mathfrak i_{f } }
h_{n^2}(d_1u(n ) \times s^1\ , | \ , p ) \ar[ur]_{\oplus ( j \times f|p)_\ast } \ar[ur ] } \ ] ] where the @xmath233-th component of the arrow @xmath394 is induced by the evident map @xmath395 and @xmath396 is as in lemma [ lem : globalint ] . by lemma [ lem :
globalint ] , the composite @xmath397 maps @xmath354 $ ] to a class which , when paired with the orientation class , yields @xmath398 . by proposition [ prop : mult_in_hom ] ,
proposition [ prop : losum ] , and remark [ rem : add - to - prop ] , the composite @xmath399 maps @xmath354 $ ] to a class which pairs to @xmath400 .
invoking the commutativity of the diagram completes the proof .
recall the scattering matrix @xmath401 defined in [ sec : intro ] .
the _ winding number _
@xmath402 of a vertex @xmath403 is the degree of the composition @xmath404 the following proposition identifies the global intersection index in terms of the winding numbers and the length function @xmath405 .
[ prop : glwind ] @xmath406 by poincar duality , the class @xmath407 ) \in h_1(u(n))$ ] is given by @xmath408 , where @xmath409 is the generator which hits the fundamental class of @xmath385 under @xmath410 .
the result now follows from the calculation @xmath411 = \text{\rm det}_\ast(\alpha_\gamma\iota ) = \alpha_\gamma\text{\rm det}_\ast(\iota ) = \alpha_\gamma [ s^1]\ , .\ ] ] the displayed formula involves a computation of the determinant of @xmath127 . for @xmath368 ,
the homomorphism property of the determinant implies the equation @xmath412 the number @xmath413 is therefore just the degree of the map @xmath414 , and is given by the sum of the degrees of the two maps appearing on the right side of eqn . .
since @xmath415 is a diagonal operator , the term @xmath416 is given by @xmath417 .
its degree is therefore the sum @xmath418 .
the determinant of @xmath419 is the product of @xmath420 indexed over the vertices of @xmath52 , so its degree is @xmath421 .
it follows that the degree of @xmath422 is given by the expression @xmath423 by the inequality in conjunction with theorem [ bigthm : indexthm ] , we infer @xmath424 the result now follows by proposition [ prop : glwind ] .
let @xmath425 be a point for which @xmath426 has an eigenvalue equal to + 1 .
set @xmath427 set @xmath428 .
then , taking the first taylor approximation of @xmath429 at @xmath430 , we obtain @xmath431 for @xmath432 and for some @xmath433 sufficiently small .
observe that the first term @xmath434 is length independent ( since @xmath435 .
the third term , which contains @xmath436 , measures the dependence of scattering on @xmath265 , which is a local phenomenon .
it too is length independent . on the other hand ,
the second term @xmath437 is length dependent .
when the edge lengths are large , the second term of the approximation dominates the third .
hence , we omit the third term of to obtain the approximation , valid up to the second order @xmath438 for @xmath439 and sufficiently large edge lengths . since
@xmath415 is given by rescaling the edge lengths , we see that each of its eigenvalues must exceed the minimal edge length , which we denote by @xmath36 , of the weighted graph . consequently , if the edge lengths are sufficiently large , the eigenvalues of the 1-jet @xmath440 are bounded below by the operator @xmath441 we infer that the imaginary part of any eigenvalue of @xmath440 is positive if @xmath442 and negative if @xmath443 for @xmath265 lying in a sufficiently small neighborhood of @xmath430 .
hence , ignoring the distinction between @xmath444 and @xmath440 , it follows by definition [ defn : iota - pm ] that @xmath445 and @xmath446 when the edge length is large .
summing up over all @xmath233 completes the proof .
wu , c. , malinin s.v .
, tretiak , s. , chernyak , v. y. : exciton scattering approach for branched conjugated molecules and complexes ii .
extraction of the exciton scattering parameters from quantum - chemical calculations .
, 174112 ( 2008 ) | this paper introduces an intersection theory problem for maps into a smooth manifold equipped with a stratification .
we investigate the problem in the special case when the target is the unitary group @xmath0 and the domain is a circle .
the first main result is an index theorem that equates a global intersection index with a finite sum of locally defined intersection indices .
the local indices are integers arising from the geometry of the stratification .
the result is used to study a well - known problem in chemical physics , namely , the problem of enumerating the electronic excitations ( excitons ) of a molecule equipped with scattering data .
we provide a lower bound for this number .
the bound is shown to be sharp in a limiting case . |
the origin of oscillatory behaviour of segmentation clock genes is believed to be due to negative feedback genetic regulation of their own products @xcite .
for example , in vertebrate developement , periodic formation of somites during presomitic mesoderm is considered to be responsible to exhibit oscillatory expression by these genes during sometogenesis and their period of oscillation is very close to the period of segmentation@xcite .
it has been reported that during this signal processing neighbouring cells are in contact and these cells communicate among the neighbouring cells showing synchronization in the dynamical variables @xcite .
there have been various theoretical studies regarding the dynamical stability , periodicity and synchronization of the various oscillating variables of segmentation clock @xcite . since synchronization and oscillation in the clock gene regulations
are believed to be responsible for somite boundary , these properties are considered to be essential in segmentation process @xcite . if a group of such cells are considered , signal processing among the neighbouring cells are done via delta - notch signaling process to achieve synchronization among them @xcite . however , since the delta proteins can not diffuse from one cell to another , the incorporation of delta - notch coupling in the theoretical models so far developed is still questionable .
one attempt in this direction without considering time delay was to introduce a coupling term in the model associated with activation rate of delta - notch signaling which enable to control the fluctuation in delta protein concentration in its neibouring cells @xcite .
it has been reported that the cells in this coupling mechanism exhibit phase synchrony in one and two dimensional array of such cells @xcite . in this work we focuss on some issues
namely , measure of phase synchrony among the cells and generalization of coupling mechanism into some more real situation . in the section
ii , we study the model of segmentation clock gene expression due to uriu et.al .
we generalized the model in three dimensional situation with coupling mechanism .
then we explain various methods to identify phase synchrony among the coupled cells .
the results and discussions of our simulation results are presented in section iii .
we draw some conclusions based on our results in section iv .
[ fig1 ] [ fig2 ]
in vertebrate sometogenesis , the expression of @xmath0 gene in single cell model with and without time delay is essential in order to generate oscillatory behavior of the segmentation clock genes @xcite . here , we consider a model in which simple kinetics of mrna , protein in cytoplasm and protein in nucleus are considered without taking into account any time delay @xcite .
the detail molecular reaction pathways are shown in fig . 1 . [ fig3 ] , @xmath1 and @xmath2 for @xmath3 and 0.7 .
the other values of the parameters used in the simulation are ; @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 , @xmath9 , @xmath10 , @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17,@xmath18 @xcite.,height=170 ] in this model @xcite , if @xmath19 , @xmath20 , @xmath21 and @xmath22 indicate concentrations of @xmath23 mrna , @xmath23 protein in cytoplasm , @xmath23 protein in nucleus and delta protein respectively in a single cell , then the time evolution of these variables without considering time delay in describing the kinetics in the cell are given by , @xmath24 where , @xmath25 , @xmath26 , @xmath27 , @xmath28 , @xmath29 and @xmath30 .
the parameters in the differential equations are given as follows : @xmath31 is the basal transcription rate , @xmath32 indicates the activation rate due to delta - notch signaling , @xmath33 is maximum degradation rate of @xmath23 mrna , @xmath34 refers to translation rate of her protein , @xmath35 is the maximum degradation rate of @xmath23 protein in cytoplasm , @xmath36 indicates transportation rate of @xmath23 protein , @xmath37 is maximum degradation rate of @xmath23 protein in nucleus , @xmath38 is the sysnthesis rate of delta protein , @xmath39 is the maximum degradation rate of delta protein , @xmath40 is threshold constant of supression of @xmath23 mrna transcripted by @xmath23 protein , @xmath41 is michaelis constant for @xmath23 mrna degradation , @xmath42 is michaelis constant for @xmath23 protein degradation in cytoplasm , @xmath43 is michaelis constant for @xmath23 protein degradation in nucleus , @xmath44 is threshold constant for the suppression of delta protein synthesis by @xmath23 protein , @xmath45 is michaelis constant for delta protein degradation in nucleus .
a group of such cells do communicate via delta - notch coupling mechanism when the cells are very close to each other or almost in contact each other .
since delta - proteins are confined only on the cell membrane and can not diffuse from one cell to another , signal is considered to pass from one cell to another by activating the delta - notch signaling pathway when the cells are in contact or very close .
we consider @xmath46 cells which are arranged in a regular three dimensional array in this type of communication , assuming the sizes of the cells are the same as shown in fig .
2 . in this situation
, any cell is allowed to couple with six nearest neighbour cells .
since the coupling variable @xmath22 is common to @xmath46 identical cells , this species can be an effective means of coupling each cell to every other via @xmath47 , where , @xmath48 and summation is restricted to nearest neighbours only for every cells .
the common species provides a `` localized '' mean field like coupling whereby the cells communicate , and the remaining variables synchronize .
[ fig4 ] , @xmath1 , @xmath2 as a function of time .
the same parameters are used as in fig .
3 . ] [ fig5 ] for one and two dimensional array , the allowed number of coupled cells are two and four respectively .
so for the cells in this configuration , the delta - notch signaling activation rate , @xmath32 plays switching on or off of signal processing among the cells and in `` on switch '' condition , coupling in any cell is done via coupling of its nearest neighbour cells . therefore in delta - notch signaling process
, we have the following coupled oscillators , @xmath49 where , @xmath50 , @xmath51 for 1 , 2 , 3 dimensional array of cells and @xmath52 , @xmath53 , @xmath54 indicate that the pairs @xmath55 , @xmath56 and @xmath57 respectively are nearest neighbours .
the mode of coupling among the cells is quite similar to mean field coupling scheme @xcite but the way how the coupling is incorporated among the cells in this case is localized within the nearest neighbours only . in this coupling mechanism , once the coupling is switch on with a particular value of @xmath32 , then the cells start synchronized among themselves to exhibit coordinated behaviour . in our simulation
we use fixed boundary conditions in the coupling , i.e. @xmath58 , @xmath59 , @xmath60 ; @xmath61 , @xmath62 , @xmath63 .
it has been pointed out that the identification of phase synchrony of any two identical systems can be done qualitatively by the measurement of the time evolution of instantaneous phase difference of the two systems @xcite .
it is possible if one defines an instantaneous phase of an _ arbitrary _ signal @xmath64 via hilbert transform @xcite @xmath65 where @xmath66 denotes the cauchy principal value .
then , one can determine an instantaneous `` phase '' @xmath67 and an instantaneous `` amplitude '' @xmath68 of the given signal through the relation , @xmath69 . given two signals of two systems @xmath70 and @xmath71 , one can therefore obtain instantaneous phases @xmath72 and @xmath73 ; phase synchronization is then the condition that @xmath74 is constant , where @xmath75 and @xmath76 are integers .
of most interest are the cases @xmath77 = 0 or @xmath78 , namely the cases of in phase or anti phase , but other temporal arrangements may also occur .
the phase synchronization of the two identical systems can also be identified by doing pecora - caroll type recurrence plot of the variable of one system , say @xmath79 with the corresponding variable @xmath80 of the other system simultaneously on two dimensional cartesian co - ordinate system @xcite .
the rate of synchronization between the two systems can be determined by the rate of concentration of the points in the plot along the diagonal .
if the two systems are uncoupled , then the points on the plot will scatter away randomly from the diagonal .
[ fig6 ] the rate of phase synchrony of the two systems can be estimated quantitatively by measuring the `` phase locking value '' ( plv ) of the two signals of the two systems @xcite .
the phase locking value , which is used to quantify the degree of synchrony , characterizes the stability of phase differences between the phases @xmath72 and @xmath73 of two signals @xmath70 and @xmath71 of ith and jth systems and can be defined within a time period @xmath81 by , @xmath82 the range of plv value is [ 0,1 ] .
when the value of plv is zero , the two systems are uncoupled , whereas if the value of plv is one then the two systems are perfectly phase synchronized @xcite .
we first present one time simulation results of @xmath19 , @xmath20 and @xmath21 of a single cell as a function of time for the values of @xmath3 and 0.7 respectively in fig . 3 by solving differential equations 1,2,3 and 4 by standard 4th order runge kutta numerical integration method @xcite . in the 3d plot , the dynamics shows three different limit cycle oscillations for these three different @xmath32 values . in fig .
4 , the corresponding dynamics of the three variables for two coupling cells are shown in panels ( a ) , ( c ) and ( e ) indicating phase synchrony irrespective of variation in @xmath32 .
the claim is supported by the phase plots i.e. @xmath83 vs time in fig . 4 ( b ) , ( d ) and ( f ) , along with pecora - caroll type recurrence plots in the panels ( g ) , ( h ) and ( i ) .
it is noticed from the plot that the amplitudes of oscillations of all variables are increased as @xmath32 is increased however the frequencies of oscillations are decreased correspondingly .
the phase locking value for whole time period ( 0 - 1500 ) minutes using equation ( [ plv ] ) is found to be [email protected] indicating fairly well synchronized .
there are signatures of fluctuations in the beginning of each plot due to the cells evolve in different initial conditions and takes some time to synchronize .
this means that once the coupling is switch on via @xmath32 among the cells , the variation in @xmath32 will hardly effect in the synchrony rate of the cells but will occur at different amplitudes .
next we take @xmath85 such cells arranged in three dimensional array .
we take @xmath86 and apply fixed boundary conditions and simulate differential equations 5 , 6 , 7 , 8 and the results of @xmath19 , @xmath20 and @xmath21 as a function of time for 10 cells are shown in fig .
the coupling variable for each cell is @xmath22 via mean field like coupling mechanism restricting to nearest neighbours only and the coupling is switch on at time 300 minutes . in the plots of fig . 5 ( a ) ,
( c ) and ( e ) , three regimes i.e. desynchronized , transition and synchronized regimes are seen clearly .
the phase transition like behaviour is supported by the phase diagrams ( @xmath87 vs time ) of pairs of cells in figures ( b ) , ( d ) and ( f ) where desynchronized regime is indicated by random evolution of @xmath87 with time , transition regime is identified by weak fluctuation of @xmath87 around a constant value and synchronized regime is indicated by constant @xmath87 with time .
again this claim of phase transition is verified qualitatively by pecora - caroll recurrence plots in the upper three panels of fig .
6 . in these plots
if the cells are desynchronized , the points in the plots will spread away from the diagonal , however in the transition regime the points start concentrating towards the diagonal .
when the points lie along the diagonal the cells are synchronized as shown in the plots . fixing the configuration of cells
, we calculated @xmath87 of pair of far distant cells ( i.e. i=1 , j=10 ) as a function of time , which are already included in the respective diagrams , and found the same pattern of phase transition .
this means that once the coupling is switch on , all the cells process information almost instantaneously even at far distant cells also in the configuration .
hence long range information transfer or relay is done instantaneously when the cells are coupled . the quantitative identification of this phase transition can be explained more clearly by the measurement of phase locking values ( @xmath88 , @xmath89 and @xmath90 ) of the variables @xmath19 , @xmath20 and @xmath21 for different time intervals by using the equation ( [ plv ] ) .
the results are shown in the lower three panels of fig .
6 . in these plots , it is seen distinctly the demarcation of the three regimes , desynchronized , transition and synchronized regimes . in the synchronized regime , the values of @xmath88 , @xmath89 and @xmath90 are [email protected] starting from time interval ( 400 - 500 ) minutes to ( 1100 - 1200 ) minutes . in the time intervals before switching the coupling , the values of @xmath88 , @xmath89 and @xmath90 are in [email protected] indicating desynchronized regime .
but in the time intervals 300 to 500 minutes , the values of @xmath88 , @xmath89 and @xmath90 monotonically increased indicating transition regime .
the phase synchrony shown by temporal oscillations of the variables indicates that once the coupling is switched on with the mean field like coupling mechanism with a value of activation rate @xmath32 above some critical value to exhibit synchrony , then the increase of @xmath32 does not affect much in the rate of phase synchrony .
however it affects in the magnitudes of amplitudes and frequencies of oscillations of the dynamical variables of the cells .
it is also to be noted that if the variables evolve with time from different initial conditions , after coupling is switched on , it takes some time to get synchronized .
since the cells coupled only when they are very close or in contact and the delta protein can not diffuse from one cell to another , the mean field like coupling scheme is the most reasonable way of signal processing in this situation .
we take three dimensional array of cells to see the behaviour of phase synchrony in more real situation .
our results interestingly show phase transition like behaviour by identifying desynchronized , transition and synchronized regimes .
relay information transfer among the far cells is done among the cells themselves once the coupling is introduced in a group of cells . to investigate it in more real situation one has to incorporate noise fluctuations arising from internal molecular events and external environmental fluctuations .
this work is financially supported by department of science and technology ( dst ) and carried out in centre for interdesciplinary research in basic sciences , jamia millia islamia , new delhi , india .
99 f. giudicelli , e.m .
ozbudak , g.j .
wright and j. lewis , plos .
biol * 5*:e150 ( 2007 ) .
h. hirata s. yoshiura , t. ohtsuka , y. bessho , t. harada , k. yoshikawa and r. kageyama , science * 298 * , 840 - 843 ( 2002 ) .
holley , d. julich , g.j .
rauch , r. geisler and c. nusslein - volhardt , developement * 129 * , 1175 - 1183 ( 2002 ) .
y. bessho , g. miyoshi , r. sakata , r. kageyama , genes cells * 6 * , 175 - 185 ( 2001 ) s.a .
holley , r. geisler , c. nusslein - volhardt , genes dev . *
14 * , 1678 - 1690 ( 2000 ) . c. jouve , i. palmeirim , d. henrique , j. beckers , a. gossler , d. ish - horowicz and o. pourquie , developement * 127 * , 1421 - 1429 ( 2000 ) .
a.c . oates and r.k .
ho , developement * 129 * , 2929 - 2946 ( 2002 ) . i. palmeirim , d. henrique , d.
ish - horowics and o. pourquie , cell * 91 * , 639 - 648 ( 1997 ) . m. maroto , j.k .
dale , m.l .
dequeant , a.c .
petit and o. pourquie , int .
149 * , 309 - 315 ( 2005 ) .
y. masamizu , t. ohtsuka , y. takashima , h. nagahara , y. takenaka , k. yoshikawa , h. okamura and r. kageyama , proc .
sci . * 103 * , 1313 - 1318 ( 2006 ) .
j. cooke and e.c .
zeeman , j. theor .
biol . * 58 * , 455 - 476 ( 1976 ) .
baker , s. schnell and p.k .
maini , j. math . biol . * 52 * , 458 - 482 ( 2006 ) .
o. cinquin , plos comput .
biol . * 3 * , e32 ( 2007 ) .
a. golbeter and o. pourquie , j. theor . biol . *
252 * , 574 - 585 ( 2008 ) .
k. uriu , y. morishita and y. iwasa , j. theor . biol . * 257 * , 385 - 396 ( 2009 ) .
k. horikawa , k. ishimatsu , e. yoshimoto , s. kondo and h. takeda , nature * 441 * , 719 - 723 ( 2006 ) . j. lewis , curr . biol .
* 13 * , 1398 - 1408 ( 2003 ) .
ozbudak and j. lewis , plos genet * 4 * : e15 ( 2008 ) .
k. uriu , y. morishita and y. iwasa , proc .
107 * , 4979 - 4984 ( 2010 ) .
k. uriu , y. morishita and y. iwasa , j. theor . biol . * 257 * , 385 - 396 ( 2009 )
. k. uriu , y. morishita and y. iwasa , j. mth .
doi 10.1007/s00285 - 009 - 0296 - 1 .
m. barrio , k. burrage a. leier and t. titan , plos comput biol * 2 * , e117 ( 2006 ) .
hb tiedemann , e. schneltzer , s. zeiser , i. rubio - aliaga , w. wurst , j. beckers , gkh przemeck and ma hrabede , j. theor biol . * 248 * , 120 - 129 ( 2007 ) .
d. gonze , j. halloy , jc .
leloup , and a.goldbeter , c. r. biol . ,
* 326 * , 189 ( 2003 ) .
m. g. rosenblum and a. s. pikovsky , phys .
* 92 * , 114102 ( 2004 ) .
a. pikovsky , m. rosenblum and j. kurths , _ synchronization : a universal concept in nonlinear science _ ( cambridge university press , cambridge , 2001 ) .
m. g. rosenblum , a. s. pikovsky and j. kurths , phys .
* 76 * , 1804 ( 1996 ) .
a. nandi , santhosh g. , r. k. b. singh and r. ramaswamy , phys .
e , * 76 * , 041136 ( 2007 ) .
h. sakaguchi and y. kuramoto , prog .
* 76 * , 576 ( 1986 ) . l. m. pecora and t. l. caroll , phys .
* 64 * , 821 ( 1990 ) .
jp lachaux , e. rodriguez , j. martinerie and f.j .
varela , human brain map .
* 8 * , 194 - 208 ( 1999 ) .
w.h . press ,
teukolsky , w.t .
vetterling and b.p .
flannery , numerical recipe in fortran , cambridge university press , 1992 . | * abstract * the temporal behaviour of segmentation clock oscillations show phase synchrony via mean field like coupling of delta protein restricting to nearest neighbours only , in a configuration of cells arranged in a regular three dimensional array .
we found the increase of amplitudes of oscillating dynamical variables of the cells as the activation rate of delta - notch signaling is increased , however , the frequencies of oscillations are decreased correspondingly .
our results show the phase transition from desynchronized to synchronized behaviour by identifying three regimes , namely , desynchronized , transition and synchronized regimes supported by various qualitative and quantitative measurements .
+ + _ _ k__eywords : delta - notch coupling , phase synchrony , hilbert transform , phase locking value , mean field coupling . |
in the standard model of electroweak interactions @xcite , the quark mass eigenstates are related to the weak eigenstates via the unitary @xmath13 cabibbo - kobayashi - maskawa ( ckm ) matrix @xmath14 @xcite .
the nine elements in this matrix , @xmath15 , where @xmath16 and @xmath17 , are completely determined from three angles and a phase .
a nonzero phase gives violation in the weak interaction .
measurements of decays of hadrons containing @xmath2 quarks ( @xmath6 hadrons ) are of great interest because they determine the magnitudes of five of the nine elements of @xmath14 as well as the phase .
the unitarity of @xmath14 leads to nine unitarity relationships , one of which is of particular interest : @xmath18 in the complex - plane , this sum of three complex numbers is a triangle , commonly referred to as _ the _ unitarity triangle .
measurements of the weak decays of @xmath6 hadrons and the already known ckm matrix elements determine the magnitudes of the three sides of the unitarity triangle , and @xmath19 asymmetries in @xmath6 meson decays determine the three angles .
the primary goal of @xmath6 physics in the next decade is to measure precisely both the sides and angles of this triangle and test consistency within the standard model .
we can use several approximations to express eq .
[ eqn : ut ] in a more convenient form .
the elements @xmath20 and @xmath21 , where @xmath22 is the cabibbo angle , are well measured . although the elements @xmath23 and @xmath24 are not well measured , the theoretical expectations are that @xmath25 and @xmath26 . with these assumptions , eq .
[ eqn : ut ] becomes @xmath27 measurement of @xmath0 , the subject of this paper , directly impacts the determination of @xmath28 .
second - order weak processes transform a neutral @xmath6 meson into its antiparticle : @xmath31 , giving a probability for a @xmath1 to decay as a @xmath32 that oscillates with time .
the frequency of these oscillations is the mass difference @xmath0 between the @xmath6 mass eigenstates , which are linear combinations of the flavor eigenstates @xmath1 and @xmath32 .
the mass difference is proportional to @xmath33 , so in principle , a measurement of @xmath0 determines this product of ckm matrix elements . in practice , however , large theoretical uncertainties limit the precision of @xmath29 .
the same problem exists in determining @xmath30 from @xmath34 , the frequency of @xmath35 oscillations .
these theoretical uncertainties are reduced in determining the ratio @xmath36 from @xmath37 .
unfortunately , at this time , attempts to measure @xmath34 have only led to lower limits .
the determination of @xmath34 is a key future measurement of @xmath6 hadrons , since @xmath36 determines the magnitude of one of the sides of the unitarity triangle , as expressed in eq .
[ eqn : utrev ] .
measurements of @xmath0 and @xmath34 require determining the initial flavor of the @xmath6 meson , that is , whether the @xmath6 meson contained a @xmath2 quark or a @xmath38 antiquark .
flavor determination is also crucial in the measurement of @xmath19 violation in the decays of neutral @xmath6 mesons to @xmath19 eigenstates .
this paper describes a measurement of @xmath0 using data collected by the cdf experiment from @xmath39 collisions with a center - of - mass energy of 1.8 tev produced by the fermilab tevatron .
in addition , the measurement of @xmath0 is used to demonstrate the performance of two methods ( described below ) of identifying the flavor of b hadrons in the environment of @xmath39 collisions .
these methods of flavor identification will be important in the measurement of violation in the decays of neutral @xmath6 mesons and in the study of @xmath34 @xcite .
the cdf collaboration has exploited the large @xmath2 quark cross - section at the fermilab tevatron to make several precision measurements of the properties of @xmath6 hadrons , including lifetimes and masses @xcite .
although the @xmath2 quark production cross - section at the tevatron is large , the @xmath4 inelastic cross - section is three orders of magnitude larger , so specialized triggers are required to collect large samples of b hadrons . to date , the triggers that have been utilized are based on leptons ( electrons and muons ) or dileptons .
some analyses use the semileptonic decays of @xmath6 hadrons , @xmath40 ; some use the semileptonic decays of charmed particles from @xmath6 hadron decay ( _ e.g. _ @xmath41 , followed by @xmath42 ) ; and some use leptonic @xmath43 decays ( @xmath44 , followed by @xmath45 ) .
previous measurements of @xmath0 from cdf were based on data samples collected with a dimuon trigger @xcite and single - lepton triggers ( @xmath46 ) @xcite .
the single - lepton triggers were used to partially reconstruct approximately 6,000 @xmath1 mesons via their semileptonic decays @xmath47 ( in this paper , reference to a particular decay sequence implies the charge - conjugate sequence as well ) .
the analysis reported in this paper uses the same data sample collected with this trigger , but increases the number of @xmath6 mesons by over an order of magnitude by inclusively reconstructing @xmath6 hadrons that decay semileptonically .
the inclusive reconstruction is made possible by the relatively long lifetime of @xmath6 hadrons : the decay point of the @xmath6 hadron is typically separated from the production point ( the primary vertex ) by a couple of millimeters .
the inclusive reconstruction is based on identifying this decay point by associating the trigger lepton with other charged decay products to reconstruct a secondary vertex .
the oscillation frequency @xmath0 can be found from either a time independent measurement ( that is , from the total number of @xmath1 s that decay as @xmath32 s ) or from a time dependent measurement ( that is , from the rates that a state that is pure @xmath48 at @xmath49 decays as either a @xmath50 or @xmath48 as a function of proper decay time @xmath51 ) .
the latter technique has better sensitivity and allows a simultaneous study of the tagging methods since the amplitude of the oscillation depends on the effectiveness of the tagging method .
the expected rate is @xmath52 @xmath53 where @xmath54 is the mean lifetime of the two mass eigenstates of the @xmath1 ( the difference in lifetime of these two eigenstates is very small @xcite and has been neglected ) , and @xmath51 is the proper decay time of the @xmath1 in its restframe . to measure this rate ,
we need to make three measurements : ( 1 ) the proper decay time , ( 2 ) the @xmath6 flavor at decay , and ( 3 ) the produced @xmath6 flavor .
we determine the proper time by measuring the distance from the production point to the decay point in the laboratory frame combined with an estimate of the @xmath48 momentum .
the flavor at decay is determined from the charge of the lepton , assuming it comes from semileptonic @xmath6 decay . in our measurement
@xcite of @xmath0 using @xmath47 decays , the flavor at production was identified using a same - side tagging technique based on the electric charge of particles produced in association with the @xmath48 .
this flavor tag has also been applied @xcite to a sample of @xmath55 decays to measure the cp asymmetry . to identify the flavor at production in this analysis ,
we rely on the fact that the dominant production mechanisms of @xmath2 quarks in @xmath39 collisions produce @xmath56 pairs . the flavors of the @xmath6 hadrons are assumed to be opposite at the time of production . in this paper , we identify the flavor of the other @xmath6 hadron using two techniques : the soft - lepton flavor tag ( slt ) and the jet - charge flavor tag ( jct ) .
several precise measurements of @xmath0 have been published @xcite by experiments operating on the @xmath57 and @xmath58 resonances .
the measurement of @xmath0 presented here is competitive in precision and in addition quantifies the performance of the flavor tags , which are crucial for future measurements of @xmath59 violation in the decays of @xmath6 mesons and the measurement of @xmath60 at a hadron collider .
the jet - charge flavor tag is a powerful technique in studies of neutral @xmath6 meson flavor oscillations by experiments operating on the @xmath58 resonance @xcite .
this analysis is the first application of the jet - charge flavor tag in the environment of a hadron collider .
the data sample used in this analysis was collected from 90 @xmath3 of @xmath39 collisions recorded with the cdf detector at the fermilab tevatron .
the cdf detector is described in detail elsewhere @xcite .
we summarize here the features of the detector that are important for this analysis .
the cdf coordinate system has the @xmath61 axis pointing along the proton momentum , with the @xmath62 axis located in the horizontal plane of the tevatron storage ring , pointing radially outward , so that the @xmath63 axis points up .
the coordinates @xmath64 , @xmath65 , and @xmath66 are the standard cylindrical coordinates .
the cdf spectrometer consists of three separate detectors for tracking charged particles : the silicon vertex detector ( svx ) , the vertex detector ( vtx ) , and the central tracking chamber ( ctc ) , which are immersed in a magnetic field of 1.4 tesla pointed along the @xmath67 axis .
the svx @xcite consists of four concentric cylinders of single - sided silicon strip detectors positioned at radii between 3 cm and 8 cm from the beam line .
the strips are oriented parallel to the beam axis and have a pitch of 60 @xmath68 m in the inner three layers and 55 @xmath68 m on the outermost layer .
the svx is surrounded by the vtx , which is used to determine the @xmath61 coordinate of the @xmath39 interaction ( the primary vertex ) . surrounding the svx and
vtx is the ctc .
the ctc is a drift chamber that is 3.2 m long with 84 layers of sense wires located between a radius of 31 cm and 133 cm .
the sense wires are organized into five axial superlayers and four stereo superlayers with a stereo angle of @xmath69 .
the momentum resolution of the spectrometer is @xmath70^{\frac{1}{2}}$ ] , where @xmath71 is the component of momentum transverse to the @xmath61 axis ( @xmath72 ) .
charged particle trajectories reconstructed in the ctc that are matched to strip - clusters in the svx have an impact parameter resolution of @xmath73 \
\mu$]m , where the impact parameter @xmath74 is the distance of closest approach of the trajectory to the beam axis in the plane perpendicular to the beam axis .
the outer 54 layers of the ctc are instrumented to record the ionization ( @xmath75 ) of charged tracks . surrounding the ctc are the central electromagnetic calorimeter ( cem ) and the central hadronic calorimeter ( cha ) .
the cem has strip chambers ( ces ) positioned at shower maximum , and a preshower detector ( cpr ) located at a depth of one radiation length .
beyond the central calorimeters lie two sets of muon detectors . to reach these two detectors , particles produced at the primary vertex with a polar angle of @xmath76
must traverse material totaling 5.4 and 8.4 pion interaction lengths , respectively .
the trigger system consists of three levels : the first two levels are implemented in hardware .
the third level consists of software reconstruction algorithms that reconstruct the data , including three - dimensional track reconstruction in the ctc using a fast algorithm that is efficient only for @xmath77 .
the sample selection begins with data from the inclusive @xmath78 and @xmath68 triggers . at level 2 ,
both of these triggers require a track with @xmath79 found by the central fast tracker ( cft ) @xcite , a hardware track processor that uses fast timing information from the ctc as input .
the resolution of the cft is @xmath80 . in the case of the electron trigger
, the cft track must be matched to a cluster in the electromagnetic calorimeter , with transverse energy @xmath81 gev , where @xmath82 , and @xmath83 is the energy of the calorimeter cluster . in the case of the muon trigger
, the cft track must be matched to a reconstructed track - segment in both sets of muon detectors . in the third level of the trigger , more stringent electron and muon selection criteria , which are similar to the selection criteria described in section [ sec : lepton_id ] , are applied .
the inclusive electron data set contains approximately 5.6 million events and the inclusive muon data set contains approximately 2.0 million events .
these data are dominated by leptons from the decay of heavy flavors ( @xmath84 and @xmath85 ) and hadrons that mimic the lepton signal .
electron candidates are identified using information from both the calorimeters and the tracking detectors .
the electron calorimeter cluster in the cem must have @xmath86 gev .
the longitudinal shower profile of this cluster is required to be consistent with an electron shower with a leakage energy from the cem into the cha of less than 4% .
the lateral shower profile of the cem cluster has to be consistent with the profile determined from test beam electrons .
a track with @xmath87 gev/@xmath88 must match the electron calorimeter cluster .
this match is based on a comparison of the track position with the calorimeter cluster position determined in the ces : the difference between the extrapolated position of the track and the position of the cluster centroid must satisfy @xmath89 cm and @xmath90 cm . to identify muons , we require a match between the extrapolated ctc track and the track segment in the muon chamber in both the @xmath64-@xmath91 and @xmath64-@xmath61 view . the uncertainty in this match is taken into account and is dominated by multiple scattering in the detector material .
the transverse muon momentum must satisfy @xmath92 gev/@xmath88 . finally , to ensure optimal resolution of the @xmath6 hadron decay point , the electron and muon candidate tracks have to be reconstructed in the svx detector .
further analysis of the data sample is based on the charged particle jets in the event .
charged particles ( instead of the more commonly used calorimeter clusters ) are used to form jets in order to keep the electron and muon samples as similar as possible .
these jets are found using a cone clustering algorithm .
tracks with @xmath93 are used as jet seeds .
if two seeds are within @xmath94 @xcite , the momenta of the seeds are added together to form a new seed .
after all possible seed merging , lower momentum tracks ( @xmath95 ) that are within @xmath96 of a seed are added in to form the final jets .
the trigger lepton is always associated to a jet , and below we refer to this jet as the trigger - lepton jet .
a jet can consist of a single track with @xmath97 . in order to reconstruct the time of decay in the @xmath6 rest frame ( the proper time )
, we must measure the point of decay with respect to the primary interaction in the lab and estimate the momentum of the @xmath6 .
since the svx provides only coordinates in the plane transverse to the beam axis , the measurement of the separation between the point of decay and the primary vertex is done only in the @xmath62-@xmath63 plane .
we refer to this separation as the decay length @xmath98 .
only the component of the @xmath6 momentum transverse to the beam axis ( @xmath99 ) is needed to calculate the proper time at decay , since the decay length is measured in the @xmath62-@xmath63 plane .
the positions of the @xmath4 interactions or `` primary vertices '' are distributed along the beam direction according to a gaussian with a width of @xmath100 cm . in the plane transverse to the beam axis , these interactions follow a distribution that is a gaussian with a width of @xmath101 m in both the @xmath62 and @xmath63 dimensions . to reconstruct the primary event vertex ,
we first identify its @xmath61-position using the tracks reconstructed in the vtx detector . when projected back to the beam axis , these tracks determine the longitudinal location of the primary interaction with a precision of about 0.2 cm along the beam direction .
if there is more than one reconstructed primary vertex in an event , the trigger lepton is associated with the primary vertex closest in @xmath61 to the intercept of the trigger lepton with the beam line .
the transverse position of the primary vertex is determined for each event by a weighted fit of all tracks with a @xmath61 coordinate within 5 cm of the @xmath61-vertex position of the primary vertex associated with the trigger lepton .
the tracks used in this fit are required to have been reconstructed in the svx detector .
first all tracks are forced to originate from a common vertex .
the position of this vertex is constrained by the transverse beam envelope described above .
tracks that have large impact parameters with respect to this vertex are removed , and the fit is repeated .
this procedure is iterated until all tracks are within the required impact parameter requirement .
at least five tracks must be used in the determination of the transverse position of the primary vertex or we use the nominal beam - line position .
the primary vertex coordinates transverse to the beam direction have an uncertainty in the range 10 - 35 @xmath68 m , depending on the number of tracks and the event topology .
the reconstruction of the @xmath6 decay point ( referred to below as the secondary vertex ) in the trigger - lepton jet is based on the technique developed to identify jets formed by @xmath2 quarks coming from @xmath51 quark decay @xcite . some modifications to this technique
were necessary to maintain good efficiency for reconstructing the @xmath6 hadron decay point in our data sample , since the @xmath6 hadrons in this sample have substantially lower @xmath71 than the @xmath6 hadrons from top quark decay .
the search for a secondary vertex in the trigger - lepton jet is a two stage process . in both stages ,
tracks in the jet are selected for reconstruction of a secondary vertex based on the significance of their impact parameter with respect to the primary vertex , @xmath102 , where @xmath103 is the estimate of the uncertainty on @xmath104 .
the uncertainty @xmath103 includes contributions from both the primary vertex and the track parameters .
the first stage requires at least three candidate tracks for the reconstruction of the secondary vertex . the trigger lepton is always included as a candidate , whether or not it satisfies the @xmath102 requirement .
tracks consistent with coming from the decay @xmath105 or @xmath106 are not used as candidate tracks .
two candidate tracks are constrained to pass through the same space point to form a seed vertex .
if at least one additional candidate track is consistent with intersecting this seed vertex , then the seed vertex is used as the secondary vertex .
if the first stage is not successful in finding a secondary vertex , the second stage is attempted .
more stringent track requirements ( on @xmath102 and @xmath71 , for example ) are imposed on the candidate tracks .
all candidate tracks satisfying these stricter criteria are constrained to pass through the same space point to form a seed vertex .
this vertex has an associated @xmath107 .
candidate tracks that contribute too much to the @xmath107 are removed , and a new seed vertex is formed .
this procedure is iterated until a seed vertex remains that has at least two associated tracks and an acceptable value of @xmath107 .
the trigger lepton is one of the tracks used to determine the trigger - lepton jet secondary vertex in 96% of the events .
the decay length of the secondary vertex @xmath98 is the projection of the two - dimensional vector pointing from the primary vertex to the secondary vertex on the jet axis ( defined by the sum of all the momenta of the tracks included in the jet ) ; if the cosine of the angle between these two vectors is positive ( negative ) , then @xmath98 is positive ( negative ) .
secondary vertices from the decay of @xmath6 hadrons are expected to have positive @xmath98 , while negative @xmath98 vertices usually result from random combinations of mismeasured tracks . to reduce the background from these false vertices , we require @xmath108 , where @xmath109 is the estimated uncertainty on @xmath98 .
we require a secondary vertex to be associated with the trigger - lepton jet .
this requirement leaves us with 243,800 events : 114,665 from the electron data sample and 129,135 from the muon data sample .
the fraction of events with @xmath110 is 4.5% in the electron data sample and 5.7% in the muon - trigger sample . for reasons we discuss later
, only events with @xmath111 are used in the determination of @xmath112 and the study of the performance of the flavor tags .
the distribution of the number of jets with total transverse momentum @xmath113
gev/@xmath88 is shown in the upper plot of figure [ fig : jets ] .
approximately 60% of the events contain a second jet in addition to the trigger - lepton jet .
the lower plot in figure [ fig : jets ] shows the difference in azimuth between the trigger - lepton jet and the jet with the highest @xmath71 in these events .
a large fraction of these jets are back - to - back with the trigger - lepton jet as expected from the lowest - order processes that produce @xmath114 pairs .
we search for secondary vertices in the other jets in the events as well .
if an additional secondary vertex is found in one of these other jets , we classify this event as a `` double - vertex '' event .
if only the single secondary vertex associated with the trigger - lepton jet is found , this event is classified as a `` single - vertex '' event .
the distinction between single - vertex and double - vertex events is important in applying the jet - charge flavor tag as described below .
the next step in the analysis is to identify the flavor at production of the @xmath6 hadron that produced the trigger lepton .
we accomplish this by identifying the flavor of the other @xmath6 hadron produced in the collision .
we refer to this other @xmath6 hadron as the `` opposite @xmath6 '' in the text below .
we first search for an additional lepton coming from the semileptonic decay of this opposite @xmath6 . because the @xmath71 of this lepton is not biased by the trigger ,
it is typically much smaller than the @xmath71 of the trigger lepton , so we call this method of flavor identification the `` soft - lepton tag '' or slt
. the soft lepton can be either an electron or a muon .
the lepton selection criteria are similar to the selection criteria described in section [ sec : lepton_id ] and reference @xcite , with additional selection criteria that use @xmath75 , and pulse height in the cpr and ces .
the soft lepton must have a track with @xmath115 gev/@xmath88 , and the invariant mass of the soft lepton and the trigger lepton must be greater than 5 gev/@xmath116 .
this requirement removes soft leptons coming from sequential semileptonic decays of charm particles produced in the decay of the @xmath6 hadron producing the trigger lepton .
the lepton identification criteria restrict the acceptance of the soft leptons to a pseudorapidity of @xmath117 , where .
approximately 5.2% of the 243,800 events contain a soft lepton candidate .
if a soft lepton is not found , we try to identify the jet produced by the opposite @xmath6 .
we calculate a quantity called the jet charge @xmath118 of this jet : @xmath119 where @xmath120 and @xmath121 are the charge and momentum of the @xmath122 track in the jet and @xmath123 is a unit - vector defining the jet axis . for @xmath2-quark jets ,
the sign of the jet charge is on average the same as the sign of the @xmath2-quark that produced the jet , so the sign of the jet charge may be used to identify the flavor at production of the @xmath6 hadron producing the trigger lepton .
if a second jet in the event other than the trigger - lepton jet has a secondary vertex , then we use this jet to calculate the jet charge .
double - vertex events with a jet - charge flavor tag are referred to as jcdv events .
if only the trigger - lepton jet contains a secondary vertex , we search for a jet with @xmath124 with respect to the trigger lepton and @xmath113 . if there is more than one jet satisfying the above criteria , we choose the jet with the highest @xmath71 .
single - vertex events with a jet - charge flavor tag are referred to as jcsv events .
approximately 7.5% of the 243,800 events above are jcdv events and approximately 42% are jcsv events .
the events in our selected data sample come from three sources : @xmath114 production , @xmath125 production , and light quark or gluon production . in each event
, the trigger lepton may be a true lepton or it may be a hadron that mimics the experimental signature of a lepton ( a fake lepton ) .
the secondary vertex in the trigger - lepton jet may be a true vertex due to the decay of heavy flavor ( @xmath2 or @xmath88 ) or a random combination of erroneously reconstructed tracks that appear to form a vertex that is displaced from the primary interaction ( a fake vertex ) .
light quark or gluon jets produce false vertices with @xmath111 with equal probability as false vertices with @xmath110 . the small fraction ( @xmath1266% ) of events with @xmath110 indicates that this background is small . in the analysis
, we assume that all events come from heavy flavor ( @xmath114 and @xmath125 ) production .
the probability of a light quark or gluon event producing a fake vertex and a fake lepton is negligible , although in the evaluation of the systematic uncertainties , we take into account the possible effects of a small amount of non - heavy flavor background .
below , we describe how we determine the fraction of our samples that are due to @xmath114 production , @xmath125 production , and fake leptons .
to understand the composition of our data , we use monte carlo samples of @xmath114 and @xmath125 production .
version 5.6 of the pythia @xcite monte carlo generator was used to generate high - statistics @xmath114 and @xmath125 samples .
the @xmath114 and @xmath125 pairs are generated through processes of order up to @xmath127 such as @xmath128 and @xmath129 , where @xmath130 @xmath2 or @xmath88 .
processes of order @xmath131 , such as gluon splitting , where @xmath132 is followed by @xmath133 , are not included , but initial and final state gluon radiation is included .
the @xmath2 and @xmath88 quarks are hadronized using the peterson @xcite fragmentation function with the parameters @xmath134 and @xmath135 .
the bottom and charm hadrons were decayed using version 9.1 of the cleo monte carlo qq @xcite .
events with a lepton with @xmath136 gev/@xmath88 were accepted based on an efficiency parameterization @xcite of the cft trigger that depends on the lepton @xmath71 .
the accepted events were passed through a simulation of the cdf detector that is based on parameterizations and simple models of the detector response that are functions of the particle kinematics . after the simulation of the cdf detector , the monte carlo events were treated as if they were real data .
the trigger - electrons in the sample can come from three sources : heavy flavor decay ( @xmath137 , @xmath138 , and @xmath139 ) , photon conversion ( @xmath140 or @xmath141 ) , or hadrons that fake the electron signature in the detector .
the contribution from heavy flavor decay is discussed in section [ sec : heavy_flavor ] .
we attempt to identify and reject photon conversions by searching for the partner of the trigger - electron .
we search for an oppositely charged track that forms a good , zero opening angle vertex with the trigger - electron .
the @xmath75 of this track , as measured in the ctc , must be consistent with the electron hypothesis .
we removed 2% of the electron trigger sample that was identified as photon conversions .
we estimate that about 1% of the remaining events contain a trigger - electron from a photon conversion that was not identified . to determine the fraction of events that contain a hadron that fakes an electron , we fit the trigger - electron @xmath75 spectrum for its @xmath78 , @xmath142 , @xmath143 , and @xmath144 content .
we found the non - electron fraction of the sample to be ( 0.6 @xmath145 0.5)% , where the uncertainty is statistical only .
since this background is small , we neglect it in the remainder of the analysis .
the trigger muons in the sample can come from heavy flavor decay ( @xmath146 , @xmath147 , and @xmath148 ) , @xmath142 and @xmath143 decay , and from hadrons that penetrate the absorbing material in front of the muon chambers .
the contribution from heavy flavor decay is discussed in section [ sec : heavy_flavor ] . to study the properties of fake muon events
, we used a control sample of events that only required a high @xmath71 cft track in the trigger .
the trigger track was treated like a trigger lepton , and the jet containing the trigger track was required to contain a secondary vertex .
the @xmath98 distribution of the control sample was very similar to the heavy flavor monte carlo @xmath98 distributions and the @xmath98 of the signal data samples .
we conclude from this comparison that due to the secondary vertex requirement most of the fake muon events in the data are events from heavy flavor production .
as described in appendix [ app : fake_muons ] , we estimate the fraction of events with a fake trigger muon whose dilution is zero by comparing the flavor tagging performance of the @xmath78 and @xmath68 trigger data .
the estimated fraction of events with fake muons that have zero dilution is @xmath149% .
we determine the fraction of events in the data due to @xmath114 and @xmath125 production using two kinematic quantities : the trigger - lepton @xmath150 and the invariant mass @xmath151 of the cluster of secondary vertex tracks .
the quantity @xmath150 is defined as the magnitude of the component of the trigger - lepton momentum that is perpendicular to the axis of the trigger - lepton jet .
the trigger lepton is removed from the jet , and the jet axis is recalculated to determine @xmath150 . to calculate @xmath151 , we assign the pion mass to all of the tracks used to form the secondary vertex ( except the trigger lepton ) .
we include the trigger lepton even if it is not attached to the secondary vertex .
these kinematic quantities are effective in discriminating between @xmath114 and @xmath125 events because of the significant mass difference between the hadrons containing @xmath2 and @xmath88 quarks ( @xmath152 gev/@xmath116 ) .
template @xmath150 and @xmath151 distributions were obtained from the @xmath114 and @xmath125 monte carlo samples .
the @xmath150 and @xmath151 distributions for the data were fit to the sum of the @xmath114 and @xmath125 monte carlo templates , where the normalization for each template was a free parameter .
the @xmath78 and @xmath68 trigger data were fit separately , and the data for each trigger were divided according to flavor tag .
the three categories were : soft - lepton ( slt ) , jet - charge single - vertex ( jcsv ) , and jet - charge double - vertex ( jcdv ) .
the results of the @xmath151 and @xmath150 fits were averaged to obtain the nominal values for the fraction of events from @xmath114 production ( @xmath153 ) .
the fits are shown in fig .
[ fig : samp_comp_ptrel ] and fig .
[ fig : samp_comp_mcl ] , and table [ tab : mclptrel ] gives the nominal values of @xmath153 for the @xmath78 and @xmath68 trigger data .
the data are mostly ( @xmath154 ) from @xmath114 production .
as outlined in the introduction , to measure @xmath112 we compare the flavor of the @xmath48 meson when it was produced to the flavor of the @xmath48 meson when it decays as a function of the proper decay time of the meson . to reconstruct the time of decay in the @xmath6 rest frame ( @xmath51 ) , we must combine the two - dimensional decay length ( @xmath98 ) with the component of the @xmath6 momentum in the @xmath62-@xmath63 plane ( @xmath99 ) .
the proper time is @xmath155 where @xmath156 is the mass of the @xmath48 and @xmath88 is the speed of light .
the proper decay length is the proper time multiplied by the speed of light ( @xmath157 ) .
we do not observe all of the decay products of the @xmath6 : the neutrino from the semileptonic decay is not detected , as well as other neutral decay products and charged decay products that may not have been associated with the secondary vertex .
this means that @xmath99 is not known and must be estimated based on observed quantities and the @xmath114 monte carlo . the momentum in the transverse plane of the cluster of secondary vertex tracks @xmath158 and @xmath151 are the observed quantities used in the estimation of @xmath99 .
the trigger lepton is included in the calculation of @xmath158 and @xmath151 , even if it is not attached to the secondary vertex , since we assume that it is a @xmath6 decay product .
the estimate of the @xmath6 hadron transverse momentum @xmath159 for an event is @xmath160 where @xmath161 is the mean of the distribution of @xmath162 determined with the @xmath114 monte carlo .
figure [ fig : kfac ] shows @xmath143 distributions for two ranges of @xmath151 and two ranges of @xmath158 in the @xmath78-trigger @xmath114 monte carlo sample . for higher @xmath151 and @xmath158 values , the @xmath143 distribution has a higher mean and a narrower width : a larger fraction of the @xmath6 momentum is observed so the observed @xmath158 is a more precise estimate of the @xmath6 momentum . to take advantage of the @xmath151 and @xmath158 dependence of the @xmath143 distribution ,
we bin the data in four ranges of @xmath151 and @xmath158 for a total of 16 @xmath143 distributions . different sets of @xmath143 distributions are used for the @xmath78 and @xmath68 trigger data .
figure [ fig : ctau ] shows the reconstructed proper decay length ( @xmath157 ) distributions for the @xmath78 and @xmath68 trigger data .
the plots of the data are compared to the expected shape from @xmath114 and @xmath125 production , where the @xmath114 and @xmath125 distributions were combined using @xmath153 in table [ tab : mclptrel ] and setting the fraction of @xmath125 events to @xmath163 .
the lack of events near @xmath164 is due to the transverse decay length significance requirement @xmath165 .
the exponential fall - off of the data in @xmath157 agrees with the monte carlo prediction .
there is , however , an excess of events with @xmath166 in the data .
this excess is due in part to backgrounds and higher - order processes not included in the monte carlo .
these backgrounds include electrons from photon conversions and fake muons .
the higher - order processes include gluon splitting to @xmath56 pairs , which can produce a pair of @xmath6 hadrons that are close in @xmath167 . in these events , the decay products of both @xmath6 hadrons
may be included in the same jet .
this leads to secondary vertices that include tracks from both @xmath6 hadrons , resulting in a specious measurement of @xmath98 that may be positive or negative . to verify this
, we generated a sample of @xmath56 events using the isajet @xcite event generator , which includes gluon splitting .
this sample showed an increased fraction of reconstructed vertices with negative @xmath98 .
the flavor of the @xmath6 hadron associated with the trigger lepton at the time of decay is identified by the trigger - lepton charge , assuming the lepton is from a semileptonic @xmath6 decay .
a @xmath6 hadron that contains an anti-@xmath2 quark ( @xmath38 ) will give a positively charged lepton in a semileptonic decay .
the trigger lepton can also originate from the decay of a charmed hadron produced in the decay of the @xmath6 hadron , _ e.g. _
@xmath168,@xmath169 , which produces `` wrong - sign '' trigger leptons , or @xmath170 , @xmath171 , which produces both wrong - sign and right - sign trigger leptons .
we refer to trigger leptons from these sources as sequential leptons .
the monte carlo @xmath114 sample is used to determine the fraction of trigger leptons , @xmath172 , from these sequential decays .
we find @xmath172 is 9.4% in the electron sample and 13.6% in the muon sample .
approximately 75% of these sequential decays , _
i.e. _ , 7.0% and 10.2% , produce wrong - sign leptons with the charge opposite to the charge from direct semileptonic decay .
we assign a systematic uncertainty of 25% of its value to the fraction of sequential leptons , based on uncertainties on the branching fractions included in the cleo monte carlo qq and on measurements of these branching fractions at the @xmath57 and the @xmath173 resonance . to determine the flavor of the trigger - lepton @xmath6 at the time of production , we attempt to identify the flavor of the other @xmath6 in the event , and assume that the original flavor of the trigger - lepton @xmath6 is opposite that of the other @xmath6 . as described previously , we use two methods to obtain the flavor of the other @xmath6 in the event : soft - lepton tagging ( slt ) and jet - charge tagging ( jct ) .
the jet - charge tag has two sub - classes : jet - charge double - vertex ( jcdv ) and jet - charge single - vertex ( jcsv ) .
the soft - lepton method is the most effective ( _ i.e. _ , has the highest probability of producing a correct tag ) , but least efficient method .
the jet - charge methods are less effective , but more efficient .
the presence of a secondary vertex in the jet used for the jet charge greatly enhances its effectiveness .
we quantify the statistical power of the flavor tagging methods with the product @xmath9 , where @xmath174 is the efficiency for applying the flavor tag , and @xmath175 is the dilution , which is related to the probability that the tag is correct ( @xmath176 ) by @xmath177 we measure @xmath9 in our data .
to illustrate the statistical significance of the product @xmath9 , we discuss an asymmetry measurement with two types of events , @xmath178 and @xmath2 , where the flavor tagging method identifies whether the event is of type @xmath178 or type @xmath2 .
type @xmath178 and type @xmath2 could be `` mixed '' and `` unmixed '' decays of a neutral @xmath6 meson , for example .
the measured asymmetry @xmath179 is @xmath180 where @xmath181 and @xmath182 are the number of events that are tagged as type @xmath178 and type @xmath2 , respectively .
the true asymmetry @xmath183 is @xmath184 where @xmath185 and @xmath186 are the true number of events of type @xmath178 and type @xmath2 , respectively , in the sample .
the efficiency is @xmath187 the true asymmetry is related to the measured asymmetry by @xmath188 and the statistical uncertainty on the true asymmetry is @xmath189 where @xmath190 is the total number of events in the sample @xmath191 .
the statistical power of different flavor tagging methods varies as @xmath9 .
we measure the dilution of the flavor tags from our data sample .
we start by defining a raw dilution , @xmath192 : @xmath193 where @xmath194 and @xmath195 are the number of right - sign and wrong - sign events , respectively .
right - sign ( wrong - sign ) means that the charge of the trigger lepton is opposite to ( the same as ) the charge of the soft lepton or jet charge .
if the charge of the trigger lepton unambiguously identified the flavor at production of the @xmath6 hadron , then @xmath192 would be equal to the true dilution @xmath175 of the flavor tag .
all wrong - sign events would result from the flavor tag being incorrect .
however , since some wrong - sign events result from the trigger lepton coming from a @xmath6 meson that mixed or from a sequential decay , @xmath192 is an underestimate of the true dilution of the flavor tag .
backgrounds from @xmath125 production and fake leptons further complicate the interpretation .
nevertheless , the true dilution is approximately related to @xmath192 by a scale factor @xmath196 : @xmath197 the form of equation [ eqn : drawtod ] is derived in appendix [ app : nd ] .
we use this estimation of the true dilution to estimate the probability that the flavor tag is correct on an event - by - event basis : @xmath198 this probability is used in the measurement of @xmath112 as described in section [ sect : fit ] .
the dilution normalization @xmath196 is determined simultaneously with @xmath112 . to maximize the effectiveness of the soft - lepton flavor tag
, the data are binned in the @xmath150 of the soft lepton .
the @xmath150 of the soft lepton is defined in the same way as the @xmath150 of the trigger lepton .
the same - sign and opposite - sign soft - lepton @xmath150 distributions are shown in figure [ fig : sltptrel ] , where the sign comparison is between the trigger - lepton charge and soft - lepton charge .
the @xmath199 bin is for the case where the soft lepton is the only track in the jet , so that @xmath200 . if neither @xmath6 decayed in a mixed state and if both leptons are from semileptonic decay , the charge of the trigger lepton would be opposite the charge of the soft lepton .
figure [ fig : sltdraw ] shows the raw dilution @xmath192 for the soft - lepton tagged events .
the raw dilution is derived from the number of same - sign and opposite - sign events in each bin of @xmath150 .
the dilution is lower for low @xmath150 because fake leptons and leptons from sequential semileptonic decay tend to have relatively low @xmath150 values .
the @xmath150 dependence of @xmath192 was parameterized using the form @xmath201 where @xmath183 and @xmath6 are parameters determined from the data .
the average raw dilution is used for isolated soft leptons that have no measurement of @xmath150 .
the form of equation [ eqn : sltd ] is empirical and was found to describe the shape of @xmath202 as a function of @xmath150 well in the monte carlo .
parameters @xmath183 and @xmath6 are measured separately for the @xmath78 and @xmath68 trigger data because the fraction of trigger leptons from sequential decay and the trigger lepton purity are different for the @xmath78 and @xmath68 data , which affects the raw dilution .
parameters @xmath183 and @xmath6 are also measured separately for the soft-@xmath78 and soft-@xmath68 tags because the fractions of fake soft leptons may be different for soft electrons and soft muons .
the fitted values of @xmath183 and @xmath6 and the average raw dilution for isolated ( no @xmath150 ) soft leptons for soft-@xmath78 and soft-@xmath68 tags in the @xmath78 and @xmath68 trigger data are listed in table [ tab : sltab ] .
the values of @xmath183 and @xmath6 determined from the data are assumed to describe the slt raw dilution as a function of @xmath150 for @xmath56 events .
this is only an approximation since a small fraction ( less than @xmath203 ) of the data are @xmath125 events .
a @xmath125 event in which @xmath204 and @xmath205 can produce opposite - sign events in which one of the leptons is associated with a secondary vertex , and the other lepton produces a soft - lepton tag . the soft - lepton tags in @xmath125 events have a much softer @xmath206 spectrum than soft - lepton tags in @xmath114 events .
the monte carlo predicts that @xmath207 events affect the values of @xmath183 and @xmath6 by an amount less than the statistical uncertainties on the fitted values of @xmath183 and @xmath6 .
the effects of the approximation above are accounted for since the values of @xmath183 and @xmath6 are varied by their statistical uncertainties in the determination of the systematic uncertainties on @xmath112 and the dilution normalization parameters .
figure [ fig : jcdist ] shows the jet charge distributions for single - vertex and double - vertex events .
the data have been divided into events with a positively or negatively - charged trigger lepton ( @xmath208 ) .
there is an anticorrelation between the sign of the jet charge and the trigger - lepton charge on average .
the degree of separation between the @xmath209 and @xmath210 distributions is related to the raw dilution of the jet - charge flavor tag , shown in figure [ fig : dplot ] as a function of the magnitude of the jet charge @xmath211 . for double - vertex events , the presence of the second
secondary vertex increases the probability that the jet selected for the calculation of the jet charge is in fact the other @xmath6 in the event .
this translates into a significantly higher raw dilution for double - vertex events .
the @xmath211 dependence of @xmath192 is used to predict the probability that the jet - charge tag is correct on an event - by - event basis , just as @xmath150 is used for soft leptons .
the @xmath211 dependence of @xmath192 in the data in figure [ fig : dplot ] was parameterized with the form @xmath212 excluding events with @xmath213 ( the rightmost data point ) . for events with @xmath213 ,
the average @xmath213 dilution is used .
the slope @xmath214 is determined separately for the single - vertex and double - vertex events in the @xmath78 and @xmath68 trigger data , respectively .
these slopes and the average raw dilution for @xmath215 are listed in table [ tab : jcdmax ] .
we use equation [ eqn : ptag ] , with @xmath192 estimated for each event using equation [ eqn : sltd ] for slt events and equation [ eqn : jcd ] for jct events , in the determination of @xmath112 to effectively discriminate between high and low quality flavor tags .
we use an unbinned maximum likelihood fit to simultaneously determine @xmath112 and the dilution normalization @xmath196 .
the effectiveness ( @xmath9 ) of each flavor tag is derived from the measured @xmath174 , @xmath192 , and the dilution normalization ( @xmath196 ) from the fit .
the fraction of @xmath48 mesons that decay in a mixed state as a function of the proper time at decay is given by @xmath216 in a pure @xmath217 sample with perfect flavor tagging and proper time resolution , @xmath218 would be equivalent to the fraction of same - sign events , comparing the sign of the flavors at decay and production .
if the flavor tag is imperfect , equation [ eqn : fmix ] becomes @xmath219 where @xmath176 ( @xmath220 ) is the probability that the flavor tag is correct ( incorrect ) .
using the relations @xmath221 equation [ eqn : fmixp ] reduces to @xmath222 although @xmath196 and @xmath112 are correlated , the basic concept is that @xmath196 is determined from the amplitude of the oscillation in the same - sign fraction and @xmath112 is determined from the frequency . we determine @xmath112 and @xmath196 by minimizing the negative log - likelihood : @xmath223 where @xmath224 is the probability density for events that are tagged as same - sign , and @xmath225 is the probability density for events that are tagged as opposite - sign .
each event has three inputs into the likelihood : 1 .
the assignment as same - sign or opposite - sign , which is based on the comparison of the sign of the slt or jct flavor tag with the charge of the trigger lepton , 2 . the estimated probability that the slt or jct flavor tag is correct , using equation [ eqn : ptag ] , with @xmath192 from equation [ eqn : sltd ] for the slt flavor tag or from equation [ eqn : jcd ] for the jct flavor tag , and 3 .
the decay distance @xmath98 .
in addition to the above three inputs , we use @xmath158 and @xmath151 to select the @xmath143 distribution that is used in the determination of the reconstructed proper decay - distance .
the construction of the probability densities requires several parameters .
these parameters are listed in table [ tab : fitparam ] .
both @xmath226 and @xmath227 are the sum of several terms .
first they have a term for the @xmath1 signal : @xmath228 where @xmath229 and @xmath230 are given by equations [ eqn : prtag ] and [ eqn : prmistag ] , respectively , and @xmath231 and @xmath232 are given by equations [ eqn : pnomix ] and [ eqn : pmixed ] , respectively .
next there are terms for the other @xmath6 hadrons , including terms for both @xmath233 and @xmath2 baryons , as well as a term for @xmath234 .
the terms for @xmath233 are @xmath235 where @xmath236 is the lifetime of the @xmath233 , and @xmath51 is the proper decay - time .
the terms for @xmath2 baryons are similar , except that @xmath236 is replaced by the lifetime of the @xmath2 baryons , @xmath237 .
the terms for @xmath234 are similar to the terms for @xmath1 , except that @xmath112 is replaced by @xmath60 , which is assumed to be very large ( _ i.e. _ beyond our experimental sensitivity ) so that these terms effectively look like @xmath238 the values of the lifetimes of the @xmath6 hadrons and the value of @xmath60 used in the probability densities are listed in table [ tab : fitparam ] .
each term for the various @xmath6 hadrons is multiplied by the expected relative contribution , @xmath239 , @xmath240 , @xmath241 , and @xmath242 , of these various hadrons to the data sample .
the production fractions are renormalized to take into account the different semileptonic branching fractions of the various @xmath6 hadrons : the semileptonic widths are assumed to be identical , so the semileptonic branching fractions are scaled to agree with the relative lifetimes .
in addition to the terms for direct semileptonic decay , there are terms that take into account the contribution from sequential decays .
the fraction of sequential decays @xmath172 is listed in table [ tab : fitparam ] ; 75% of these sequential decays produce leptons with a sign opposite to direct semileptonic decay .
the terms that take into account the contribution from @xmath207 events are similar to the terms for @xmath233 , except that we use an assumed flavor tagging dilution and an effective lifetime @xmath243 .
the true flavor tagging dilution is not known _ a priori _ for both bottom and charm decays .
just as for @xmath6 decays , we do not consider the monte carlo reliable for predicting the jct or slt dilution for charm decays , so we use assumed values for the charm dilution and varied them by the maximum possible amount for the systematic uncertainties .
for the jct , we expect the dilution for charm decays to be worse than for bottom decays , therefore we assume @xmath244 for the jct , and vary @xmath245 from 0 to 1 in the evaluation of the systematic errors . the ratio @xmath245 is used to rescale the predicted @xmath56 dilution for the event , based on @xmath246 using equation [ eqn : jcd ] , to give the predicted @xmath207 dilution for the event .
the slt dilution for charm decays could be anything from 0 to 1 depending on the fraction of fake soft leptons in @xmath207 events .
unlike bottom decays , the slt dilution for charm decays does not fall near @xmath200 due to soft leptons from sequential decays , therefore we assume @xmath247 , independent of @xmath150 for the slt , and vary @xmath245 from 0 to 1 in the evaluation of the systematic errors .
the effective lifetime @xmath248 ps is determined from the @xmath125 monte carlo samples , where the proper time at decay was reconstructed using the @xmath143 distributions from the @xmath114 monte carlo and equation [ eqn : ctau ] .
the relative fractions of @xmath56 ( @xmath249 ) and @xmath207 ( @xmath250 ) , which can be determined from table [ tab : mclptrel ] , multiply the terms in the likelihood corresponding to @xmath56 and @xmath207 production , respectively .
finally , for the case of the @xmath68-trigger data , a term for fake muons was included with the @xmath48 lifetime and zero dilution .
the relative fraction of fake muons @xmath251 ( discussed in appendix [ app : fake_muons ] ) is listed in table [ tab : fitparam ] . in this case ,
@xmath249 and @xmath250 are scaled by @xmath252 . the probability densities
@xmath253 and @xmath254 are functions of the true proper time at decay @xmath51 .
we take into account the experimental resolution on @xmath51 by convoluting @xmath253 and @xmath254 with @xmath98 and @xmath255 resolution functions .
distributions of @xmath256 from the @xmath78 and @xmath68 trigger @xmath114 monte carlo samples are parameterized with the sum of three gaussians , a @xmath257 exponential , and a @xmath258 exponential .
the @xmath259 distributions and their parameterizations are shown in figure [ fig : lxyres ] for the @xmath78 and @xmath68 trigger data .
the @xmath143 distributions , like those shown in figure [ fig : kfac ] , are used as @xmath255 resolution functions .
the @xmath98 and @xmath255 resolution functions describe the @xmath51 smearing for @xmath114 events from direct production ( @xmath260 and @xmath261 ) .
they do not describe backgrounds such as photon conversion events ( for the @xmath78 trigger ) or events with a fake trigger muon .
they also do not describe @xmath114 events from gluon splitting ( @xmath262 followed by @xmath263 ) , which tend to have worse @xmath51 resolution due to secondary vertices that include decay products from both @xmath6 hadrons . while the total amounts of these backgrounds are reasonably small , they become more important near @xmath164 and for very large @xmath157 values ( beyond several @xmath6 lifetimes ) . for these reasons ,
we only use events with @xmath264 in the fit for @xmath112 and the dilution normalization @xmath196 .
the free parameters in the unbinned maximum likelihood fit are the mass difference @xmath112 and the dilution normalization factors @xmath196 .
there are six dilution normalization factors : 1 .
@xmath265 and @xmath266 : the dilution normalization factors for the jet - charge flavor tag in the case that the jet does not contain a secondary vertex for the electron - trigger and muon - trigger data samples , respectively .
2 . @xmath267 and @xmath268 : the dilution normalization factors for the jet - charge flavor tag in the case that the jet does contain a secondary vertex for the electron - trigger and muon - trigger data samples , respectively .
3 . @xmath269 and @xmath270 : the dilution normalization factors for the soft - lepton flavor tag for the electron - trigger and muon - trigger data samples , respectively . to determine the dilution normalization factors needed to calculate the flavor tag @xmath9 values ( see section [ sec : eds ] ) ,
the data are grouped into four subsamples : 1 .
@xmath78-trigger , soft - lepton flavor tag , 2 .
@xmath78-trigger , jet - charge flavor tag , 3 .
@xmath68-trigger , soft - lepton flavor tag , and 4 .
@xmath68-trigger , jet - charge flavor tag .
there is some overlap of events in these four subsamples as some events have both a slt and a jct .
about 20% of the events with soft - lepton tags did not pass the single - lepton level 2 trigger and came , instead , from a dilepton trigger . in order for the slt efficiency to be well defined
, we require that the slt events pass the single - lepton trigger in data subsamples 1 and 3 , which are used to determine @xmath271 , @xmath272 , and the slt efficiencies needed for calculating the slt @xmath9 .
the fit results for the individual flavor taggers are given in table [ tab : indivfits ] .
the dilution normalization factors are given for two cases : @xmath112 free to float in the fit and @xmath112 fixed to the world average @xcite ( 0.474 @xmath273 ) .
the value of @xmath112 is held fixed to the world average because @xmath112 and the @xmath196 factors are correlated .
the correlation coefficients between @xmath112 and the @xmath196 constants range from 0.55 to 0.81 .
fixing @xmath112 reduces the statistical uncertainty on the dilution normalization constants and removes any bias from statistical fluctuations that pull @xmath112 high or low .
the @xmath196 factors determined with @xmath112 fixed to the world average are used in the calculation of the flavor tag @xmath9 values .
the @xmath196 factors in table [ tab : indivfits ] are consistent with our expectations from the composition of the data ( see appendix [ app : nd ] ) .
to determine @xmath112 , we fit all four data subsamples simultaneously . for events with both a slt and a jct
, we use the slt because it has significantly higher average dilution .
the results of the simultaneous fit of the @xmath78 and @xmath68 trigger data using both flavor tagging methods are listed in table [ tab : fit_results ] .
we find @xmath274 , where the uncertainty is statistical only .
this is consistent with the world average value ( 0.464 @xmath145 0.018 @xmath273 ) @xcite . using the slt tag in
doubly flavor tagged events removes events with higher - than - average dilution from the jct .
this results in lower values of @xmath196 for the jet - charge flavor tag than found in the fits of the individual samples listed in table [ tab : indivfits ] .
figure [ fig : ssfr ] shows the fraction of same - sign events as a function of the reconstructed proper decay length @xmath157 .
the points with error bars are the data .
the curve is a representation of the results of the fit for @xmath112 and @xmath196 .
figure [ fig : ssfr ] has been included to illustrate the clear evidence of @xmath217 mixing in the data .
however , it does not contain all of the information that goes into the unbinned likelihood fit . in figure
[ fig : ssfr ] , all events are treated equally . in the fit , events are effectively weighted based on their estimated dilution . to check our fitting procedure
, we used a fast monte carlo which generated hundreds of data samples , each with the same statistics , tagging dilution , and @xmath51 resolution as the real data .
figure [ fig : toy ] shows the fit results of 400 fast monte carlo samples , representing the slt flavor tagged , @xmath78-trigger data .
the top row of plots show distributions of the fitted values of @xmath112 and @xmath196 for the 400 samples .
the arrows indicate the values of @xmath112 and @xmath196 with which the samples were generated .
the mean of each distribution is consistent with the generation value .
the middle row of plots show the distributions of the statistical uncertainty on @xmath112 and @xmath196 .
the arrows indicate the statistical uncertainty from the fit to the slt flavor tagged , @xmath78-trigger data .
the statistical uncertainties on @xmath112 and @xmath196 for the data are near the most probable values from the fast monte carlo samples .
the bottom row of plots show the distributions of the deviation of the fitted value from the generation value divided by the statistical uncertainty for @xmath112 and @xmath196 .
these distributions have a mean of zero and unit width , which confirms that @xmath112 and @xmath196 are unbiased and that the statistical uncertainty is correct . to determine the systematic uncertainty on @xmath112 , the fixed input parameters in table [ tab : fitparam ] were varied by the amounts listed in this table . to determine the systematic uncertainty associated with the dilution parameterizations , the parameters describing the @xmath192 dependence on @xmath211 and @xmath150 of the soft lepton were varied by their statistical uncertainties .
a systematic uncertainty to account for the possible presence of fake secondary vertices from non - heavy flavor backgrounds as well as heavy flavor events ( gluon splitting ) was determined using a combination of data and monte carlo .
the observed excess in the data over the combined contribution of @xmath56 and @xmath207 ( see figure [ fig : ctau ] ) was used to define the shape of this background .
this shape was included in the likelihood function with a dilution that varied from 0 to @xmath275 .
the systematic uncertainty assigned to the variation of each input parameter was the shift in @xmath112 from the fit result with the nominal values of the input parameters .
the total systematic uncertainty is the sum in quadrature of these shifts .
the same procedure was applied to the four data subsamples ( @xmath78 slt , @xmath78 jct , @xmath68 slt , @xmath68 jct ) to determine the systematic uncertainties for the dilution normalization factors .
the procedure described above was checked for the largest individual contributions to the systematic uncertainty using fast monte carlo samples .
samples generated with a variation on one of the input parameters ( e.g. @xmath276 ) were fit using the nominal parameter ( e.g. @xmath277 ) .
the average bias on the fitted values for the fast monte carlo samples was consistent with the deviation on the fitted values observed when the data are fit with a fixed parameter variation .
table [ tab : dmdsys ] lists the individual contributions to the systematic uncertainty on @xmath112 ( @xmath278 ) . the largest single contribution ( @xmath279 )
is the unknown soft - lepton flavor tag dilution for @xmath207 events .
the slt @xmath207 dilution was varied over the full possible range ( 0 to 1 ) .
if the assumed @xmath207 dilution in the fit is lower ( higher ) than its true value , the fraction of same - sign events at small @xmath51 will appear lower ( higher ) , which biases @xmath112 low ( high ) .
the second largest contribution ( @xmath280 ) is the uncertainty on the lifetime ratio of the @xmath281 and the @xmath48 mesons .
if the value of @xmath282 in the fit is higher ( lower ) than its true value , the fraction of same - sign events at short @xmath51 values will appear larger ( smaller ) , which biases @xmath112 high ( low ) . the third largest contribution ( @xmath283 )
is the uncertainty on the slt raw dilution parameterization : @xmath202 as a function of soft - lepton @xmath150 .
the size of the variation for the slt is shown by the dashed curves in figure [ fig : sltdraw ] .
the raw dilution for events with no @xmath150 measurement was independently varied by the statistical uncertainty on the raw dilution for no-@xmath150 events .
tables [ tab : ndsys_e ] and [ tab : ndsys_m ] list the contributions to the systematic uncertainty on the dilution normalization factors for the @xmath78 and @xmath68 trigger data respectively . as for the systematic uncertainty on @xmath112 ,
the largest contribution comes from the unknown dilution for @xmath207 events .
the fraction of @xmath284 events , which we assume are half same - sign and half - opposite sign , and the fraction of events where the trigger lepton is from a sequential decay both affect the assignment of same - sign events .
if these fractions are low ( high ) , more ( less ) same - sign events will be attributed to mistags , thus they are strongly coupled to the dilution normalization .
the uncertainty on the raw dilution parameterizations also has a large effect on the dilution normalization systematic uncertainty .
the measurement of the statistical power @xmath9 of the flavor tagging methods is done in two steps .
first , the raw @xmath9 is calculated using the raw dilution rather than the true dilution .
then , the raw @xmath9 is rescaled by @xmath285 , which translates the raw dilution to the true dilution . the @xmath196 factor for each flavor tagging method in the @xmath78 and @xmath68 samples
is determined in the unbinned maximum likelihood fit with @xmath112 fixed to the world average @xcite ( 0.474 @xmath273 ) .
these @xmath196 values are given in table [ tab : indivfits ] , where the first uncertainty is statistical and the second systematic .
the raw or uncorrected value of @xmath9 is obtained by summing the @xmath286 values in the bins of either @xmath150 for the slt or @xmath211 for the jct .
the efficiency for each bin is defined as the number of events in the bin divided by the total number of events before flavor tagging .
table [ tab : eds ] lists the total efficiency , raw @xmath9 , dilution normalization , and true @xmath9 for each of the flavor tagging methods . taking the average of the @xmath78 and @xmath68 trigger data
, we find @xmath9 to be ( @xmath287)% for the jct and ( @xmath288)% for the slt where the first uncertainty is statistical and the second systematic .
these @xmath9 values are about one order of magnitude lower than typical flavor tagging techniques employed on the @xmath173 resonance @xcite , however the large @xmath56 cross section in @xmath4 collisions at @xmath289 tev yields @xmath56 samples that are about one order of magnitude larger at cdf than those collected on the @xmath173 resonance .
the statistical uncertainty for our measurement of @xmath112 is still competitive with similar measurements on the @xmath173 resonance , since our smaller @xmath9 is compensated by our larger sample size .
the values of @xmath9 for these flavor tags depend on the data sample in which they are used .
in particular , during next run of the tevatron , we will collect large samples of @xmath290 ( for the precise measurement of the @xmath19 asymmetry parameter @xmath291 ) and hadronic @xmath292 decays ( for the precise determination of @xmath60 ) .
the triggers used to collect these data samples will be different from the inclusive lepton trigger used to collect the data for this analysis . as a result , the @xmath6 hadron production properties ( _ e.g. _ , @xmath293 of the @xmath6 ) are different , and this affects @xmath9 . despite these differences , the results in this paper demonstrate that both the jet - charge and the soft - lepton flavor tagging methods are viable in the environment of @xmath4 collisions .
we have measured @xmath112 using soft - lepton and jet - charge flavor tagging methods .
this is the first application of jet - charge flavor tagging in a hadron - collider environment .
the flavor at decay was inferred from the charge of the trigger lepton , which was assumed to be the product of semileptonic @xmath6 decay .
the initial flavor was inferred from the other @xmath6 in the event , either using the charge of a soft lepton or the jet charge of the other @xmath6 .
the proper - time at decay for each event was determined from a partial reconstruction of the decay vertex of the @xmath6 that produced the trigger lepton and an estimate of the @xmath6 momentum .
the value of @xmath112 was determined with an unbinned maximum likelihood fit of the same - sign and opposite - sign proper - time distributions ( comparing the sign of the trigger - lepton charge and the flavor tag ) .
the statistical power of the flavor tagging methods was measured in the unbinned maximum likelihood fit by fitting for a scale factor @xmath196 , for each of the flavor tagging methods , which is the ratio of the raw dilution and the true dilution .
we find @xmath294 , where the first uncertainty is statistical and the second systematic .
this is consistent with the world average value of @xmath295 @xcite and competitive in precision with other individual measurements of @xmath112 .
we quantify the statistical power of the flavor tagging methods with @xmath9 , which is the tagging efficiency multiplied by the square of the dilution .
we find @xmath9 to be ( @xmath287)% for the jet - charge flavor tag and ( @xmath288)% for the soft - lepton flavor tag , where the first uncertainty is statistical and the second systematic .
these @xmath9 are much lower than what has been achieved in experiments on the @xmath173 resonance , however we have demonstrated that the much higher @xmath56 cross - section at the tevatron ( @xmath4 , @xmath289 tev ) can be used to compensate for the disadvantage in @xmath9 .
the jet - charge and soft - lepton flavor tagging techniques will be important tools in the study of @xmath19 violation in the up - coming run of the tevatron .
we thank the fermilab staff and the technical staffs of the participating institutions for their vital contributions .
this work was supported by the u.s .
department of energy and national science foundation ; the italian istituto nazionale di fisica nucleare ; the ministry of education , science and culture of japan ; the natural sciences and engineering research council of canada ; the national science council of the republic of china ; the swiss national science foundation ; and the a. p. sloan foundation . 99 s. l. glashow , nucl . phys . * 22 * , 579 ( 1961 ) : + s. weinberg , phys . rev .
19 , 1264 ( 1967 ) ; + a. salam , _ elementary particle theory _
n. svartholm , almquist and wiksells , stockholm ( 1969 ) p. 367 .
n. cabibbo , phys .
* 10 * , 531 ( 1963 ) .
+ m. kobayashi and k. maskawa , prog .
phys . * 49 * , 652 ( 1973 ) .
cdf collaboration , r. blair _
et al_. the cdfii detector technical design report , fermilab - pub-96/390-e .
cdf collaboration , f. abe _ et al .
_ , phys .
d * 59 * , 032004 ( 1999 ) .
+ cdf collaboration , f. abe _ et al .
_ , phys .
d * 58 * , 112004 ( 1998 ) .
+ cdf collaboration , f. abe _ et al .
_ , phys .
d * 58 * , 092002 ( 1998 ) . + cdf collaboration , f. abe _ et al .
_ , phys .
d * 57 * , 5382 ( 1998 ) . + cdf collaboration , f. abe _ et al .
_ , phys .
d * 55 * , 1142 ( 1997 ) .
+ cdf collaboration , f. abe _ et al .
_ , phys .
* 77 * , 1945 ( 1997 ) .
+ cdf collaboration , f. abe _ et al .
_ , phys .
d * 53 * , 3496 ( 1996 ) .
cdf collaboration , f. abe _ et al .
_ , _ measurement of the @xmath296 oscillation frequency using dimuon data in @xmath4 collisions at @xmath289 tev _ , fermilab - pub-99/030-e , submitted to physical review letters .
cdf collaboration , f. abe _ et al_. , phys .
lett . , * 80 * , 2057 , ( 1998 ) ; + cdf collaboration , f. abe _ et al_. , phys .
d * 59 * , 032001 ( 1999 ) .
c. caso _ et al . _ ( the particle data group ) , the european physics journal * c3 * ( 1998 ) 1 .
cdf collaboration , f. abe _ et al_. , phys .
lett . , * 81 * , 5513 , ( 1998 ) . a summary of published measurements and their average can be found in c. caso et al , the european physical journal * c3 * , 1 ( 1998 ) .
aleph collaboration , d. buskulic _ et al .
_ , zeit .
c * 75 * , 397 ( 1997 ) ; + delphi collaboration , p. abreu _ et al .
_ , zeit .
c * 72 * , 17 ( 1996 ) ; + opal collaboration , r. akers _ et al .
_ , phys .
b * 327 * , 411 ( 1994 ) ; + sld collaboration , k. abe _ et al .
_ , phys .
lett . * 74 * , 2890 ( 1995 ) .
et al . _ ,
. instrum .
methods phys .
a * 271 * , 387 ( 1988 ) .
f. abe _ et al .
_ , phys . rev . lett .
* 74 * ( 1995 ) 2626 ; + see also f. abe _ et al .
_ , phys .
d * 50 * ( 1994 ) 2966 .
d. amidei _ et al .
methods * a350 * , 73 , ( 1994 ) ; + p. azzi _ et al .
methods * a360 * , 137 ( 1995 ) .
g. w. foster _ et al .
_ , nucl .
instrum .
methods a * 269 * , 93 ( 1988 ) .
@xmath297 where @xmath65 is the azimuthal direction with respect to the beam and @xmath298 is the pseudorapidity , which is defined as @xmath299 .
h. u. bengtsson and t. sjstrand , computer physics commun . *
43 * , 43 ( 1987 ) ; t. sjstrand , computer physics commun . * 82 * , 74 ( 1994 ) . c. peterson _
et al . _ ,
d * 27 * ( 1983 ) 105 .
p. avery , k. read , and g. trahern , cornell internal note csn-22 , 1985 ( unpublished ) .
this efficiency parameterization followed the functional form @xmath300 where @xmath301 , @xmath302 gev/@xmath88 , @xmath303 , @xmath304 gev/@xmath88 , and @xmath305 .
the efficiency curve turns on at @xmath306 , rises to 50% at @xmath307 , and has a plateau value of 93% .
f. paige and s. d. protopopescu , brookhaven national laboratory report no .
bnl 38034 , 1986 .
r. m. barnett _
et al . _ ( the particle data group ) , physical review d*54 * , 1 ( 1996 ) .
f. james , cern program library long writeup d506 .
the true dilution @xmath175 of a flavor tagging method is defined as @xmath308 where @xmath176 is the probability that the flavor tag is correct . an equivalent expression for @xmath175
is @xmath309 where @xmath310 ( @xmath311 ) is the number of correct ( incorrect ) tags in a sample of @xmath312 events .
the raw dilution is defined as @xmath313 where @xmath314 ( @xmath315 ) is the number of opposite - sign ( same - sign ) events in the sample , comparing the trigger - lepton charge with either the sign of the soft - lepton charge or the jet charge .
if the data were pure @xmath56 with no @xmath6 mixing and all of the trigger leptons were from direct @xmath6 decay , all opposite ( same ) sign events would be correctly ( incorrectly ) flavor - tagged .
that is , we would have @xmath316 , @xmath317 , and @xmath318 .
there are , however , several things in the data that break the @xmath316 and @xmath317 assumptions .
they are * * @xmath6 mixing * : if the trigger lepton is from a @xmath6 hadron that decays in a state opposite its original flavor , the trigger - lepton charge will have the `` wrong '' sign . in this case ,
events with the correct flavor tag are same - sign . *
* sequential decays * : the charge of trigger leptons from sequential @xmath6 decay ( ) is opposite that of direct @xmath6 decay . for trigger leptons that are from sequential decay , events with the correct flavor tag
are same - sign , if the trigger - lepton @xmath6 did not mix . * * @xmath207 events * : events from @xmath207 production may have a non - zero dilution that is not the same as the dilution from @xmath56 events . * * fake leptons * : the @xmath78-trigger data has essentially no fake trigger electrons
. however , about 12% of the @xmath68-trigger data have a hadron that faked a muon , whose charge is random ( see appendix [ app : fake_muons ] ) . if there were no fake leptons , the number of opposite - sign and same - sign events from @xmath56 production are given by @xmath319 \nonumber \\ & & + f^{\rm ws}_{\rm seq } \left [ \bar{\chi } ' n^{\bbbar}_t + ( 1 - \bar{\chi } ' ) n^{\bbbar}_m \right ] \\
\label{eqn : nssbb } n^{\bbbar}_{ss } & = & ( 1 - f^{\rm ws}_{\rm seq } ) \left [ \bar{\chi } ' n^{\bbbar}_t + ( 1 - \bar{\chi } ' ) n^{\bbbar}_m \right ] \nonumber \\ & & + f^{\rm ws}_{\rm seq } \left [ ( 1 - \bar{\chi } ' ) n^{\bbbar}_t + \bar{\chi } ' n^{\bbbar}_m \right ] \end{aligned}\ ] ] where @xmath320 is the fraction of trigger leptons in @xmath56 events that are from sequential decay in which the trigger - lepton charge has the `` wrong '' sign , @xmath321 is the effective , which causes @xmath321 to be larger than @xmath322 . ]
probability that the @xmath6 hadron that produced the trigger lepton decayed in a mixed state , and @xmath323 ( @xmath324 ) is the number of same - sign ( opposite - sign ) @xmath56 events . for events from @xmath207 production we have @xmath325 where @xmath326 ( @xmath327 ) is the number of same - sign ( opposite - sign ) @xmath207 events . using equations [ eqn : appdraw ] , [ eqn : nosbb ] , [ eqn : nssbb ]
, [ eqn : noscc ] , and [ eqn : nsscc ] the raw dilution can be written as @xmath328 if a fraction of the events ( @xmath329 ) have a fake trigger lepton whose charge - sign is random , these events will have a raw dilution of zero since the number of same - sign fake - lepton events will equal the number of opposite - sign fake - lepton events .
taking fake leptons into account gives @xmath330 using equation [ eqn : appd ] , we define the true flavor tagging dilution in @xmath56 and @xmath207 events as @xmath331 and @xmath332 respectively .
the fraction of events from @xmath56 and @xmath207 production are defined by @xmath333 and @xmath334 respectively . combining equations [ eqn : draw1 ] , [ eqn : dbbdef ] , [ eqn : dccdef ] , [ eqn : fbbdef ]
, [ eqn : fccdef ] gives @xmath335 \nonumber \\
d_{\rm raw } & = & \left\ { ( 1 - f^{\ : 0}_{{\rm fake }
\ \ell } ) \left [ ( 1 - 2 \bar{\chi}')(1 - 2 f^{\rm ws}_{\rm seq } ) \
f_{\bbbar } + f_{\ccbar } \
\frac{d_{\ccbar}}{d_{\bbbar } } \right ] \right\ } d_{\bbbar } \nonumber\\ d_{\rm raw } & = & \frac{1}{n_d } d_{\bbbar } \label{eqn : draw3}\end{aligned}\ ] ] where we have defined the dilution normalization factor @xmath196 as @xmath336.\ ] ] equation [ eqn : nddef ] can be used to calculate the expected values for the @xmath196 parameters . for this calculation , we will assume * @xmath337 from the monte carlo .
* @xmath338 and @xmath339 using @xmath340 and the values in table [ tab : fitparam ] . * the @xmath249 values are given in table [ tab : mclptrel ] .
we also use @xmath341 . * for the jct
, we assume @xmath342 . using the average slt dilution and the assumption that @xmath343
, we estimate @xmath344 . using the numbers above
, we find * @xmath345 .
* @xmath346 .
* @xmath347 . * @xmath348 .
as is stated in section [ sec : fakemu ] , we believe that most of the fake trigger - muons in the data are from heavy flavor decay . there may be some correlation on average between the sign fake muon charge and the @xmath6 flavor at decay , however we assume that this correlation is smaller than that of real trigger - muons .
fake muon events are divided into two groups : we determine the fraction of events with fake muons that have zero dilution in the @xmath68-trigger data by assuming that the _ true _ flavor tagging dilution is the same in the @xmath78 and @xmath68 trigger data . the raw dilution ( @xmath192 ) , which assumes all opposite - sign ( same - sign ) events are tags ( mistags ) , is different for the @xmath78 and @xmath68 trigger data for the following reasons : 1 .
the fraction of real trigger - leptons in @xmath114 events that are not from direct @xmath349 decay ( @xmath172 ) is 9.4% for the @xmath78-trigger data and
13.6% for the @xmath68-trigger data .
the fraction of events from @xmath207 production @xmath250 is slightly different ( see table [ tab : mclptrel ] ) .
we estimate that only 1% of the trigger electrons are fake , while , as shown below , about 10% of the @xmath68-trigger events contain a fake muon . we can correct @xmath192 for ( 1 ) and ( 2 ) using the equation @xmath350 where @xmath351 is the fraction of non @xmath349 decays that have the `` wrong '' sign .
the monte carlo gives @xmath352 .
the values of @xmath353 for the slt , jcsv , and jcdv flavor tagging methods in the @xmath78 and @xmath68-trigger data are given in table [ tab : dprime ] .
the weighted average of @xmath354 for the slt , jcsv , and jcdv flavor tagging methods gives 1.14 @xmath145 0.08 .
the fraction of events with fake muons that have zero dilution can be extracted using @xmath355 equation [ eqn : ffakemu ] gives @xmath356 % .
the relatively large uncertainty on @xmath357 gives a significant systematic uncertainty on the dilution normalization @xmath196 for the flavor tags ( see table [ tab : ndsys_m ] ) , however , the contribution to the systematic uncertainty on @xmath112 is relatively small . .
the fraction of events from @xmath114 production @xmath249 for the @xmath78 and @xmath68 trigger data .
the first two columns give the result for @xmath249 from fitting the @xmath150 and @xmath151 spectra to a linear combination of @xmath56 and @xmath207 templates .
the error in the first two columns is statistical only .
the last column gives the average of the @xmath150 and @xmath151 results .
the error on the average takes into account the difference in the @xmath150 and @xmath151 fit results as a systematic error .
the fraction of events from @xmath207 production is given by @xmath358 . [ cols="<,^,^,^ " , ] | we present a measurement of the mass difference @xmath0 for the @xmath1 meson and the statistical power of the @xmath2 flavor tagging methods used . the measurement uses 90 @xmath3 of data from @xmath4 collisions at @xmath5 tev collected with the cdf detector . an inclusive lepton trigger is used to collect a large sample of @xmath6 hadron semileptonic decays .
the mass difference @xmath0 is determined from the proper time dependence of the fraction of @xmath6 hadrons that undergo flavor oscillations .
the flavor at decay is inferred from the charge of the lepton from semileptonic @xmath6 decay .
the initial flavor is inferred by determining the flavor of the other @xmath6 hadron produced in the collision , either from its semileptonic decay ( soft - lepton tag ) or from its jet charge .
the measurement yields @xmath7 @xmath8 , where the first uncertainty is statistical and the second uncertainty is systematic .
the statistical powers ( @xmath9 ) of the soft - lepton and jet - charge flavor taggers are @xmath10 and @xmath11 , respectively .
0z^0 = cmti8 # 1@xmath12 f. abe , h. akimoto , a. akopian , m. g. albrow,7 s. r. amendolia , d. amidei , j. antos , s. aota , g. apollinari , t. arisawa , t. asakawa , w. ashmanskas,5 m. atac,7 p.
azzi - bacchetta , n. bacchetta , s. bagdasarov , m. w. bailey , p. de barbaro , a. barbaro - galtieri , v. e. barnes , b. a. barnett , m. barone,9 g. bauer , t. baumann , f. bedeschi , s. behrends,3 s. belforte , g. bellettini , j. bellinger , d. benjamin,6 j. bensinger,3 a. beretvas,7 j. p.
berge,7 j. berryhill,5 s. bertolucci,9 s. bettelli , b. bevensee , a. bhatti , k. biery,7 c. bigongiari , m. binkley,7 d. bisello , r. e. blair,1 c. blocker,3 k. bloom , s. blusk , a. bodek , w. bokhari , g. bolla , y. bonushkin,4 d. bortoletto , j. boudreau , a. brandl , l. breccia,2 c. bromberg , n. bruner , r. brunetti,2 e. buckley - geer,7 h. s. budd , k. burkett , g. busetto , a. byon - wagner,7 k. l. byrum,1 m. campbell , a. caner , w. carithers , d. carlsmith , j. cassada , a. castro , d. cauz , a. cerri , p. s. chang , p. t. chang , h. y. chao , j. chapman , m. -t .
cheng , m. chertok , g. chiarelli , c. n. chiou , f. chlebana,7 l. christofek , m. l. chu , s. cihangir,7 a. g. clark , m. cobal , e. cocca , m. contreras,5 j. conway , j. cooper,7 m. cordelli,9 d. costanzo , c. couyoumtzelis , d. cronin - hennessy,6 r. cropp , r. culbertson,5 d. dagenhart , t. daniels , f. dejongh,7 s. dellagnello,9 m. dellorso , r. demina,7 l. demortier , m. deninno,2 p. f. derwent,7 t. devlin , j. r. dittmann,6 s. donati , j. done , t. dorigo , n. eddy , k. einsweiler , j. e. elias,7 r. ely , e. engels , jr .
, w. erdmann,7 d. errede , s. errede , q. fan , r. g. feild , z. feng , c. ferretti , i. fiori,2 b. flaugher,7 g. w. foster,7 m. franklin , j. freeman,7 j. friedman , y. fukui , s. gadomski , s. galeotti , m. gallinaro , o. ganel , m. garcia - sciveres , a. f. garfinkel , c. gay , s. geer,7 d. w. gerdes , p. giannetti , n. giokaris , p. giromini,9 g. giusti , m. gold , a. gordon , a. t. goshaw,6 y. gotra , k. goulianos , h. grassmann , c. green , l. groer , c. grosso - pilcher,5 g. guillian , j. guimaraes da costa , r. s. guo , c. haber , e. hafen , s. r. hahn,7 r. hamilton , t. handa , r. handler , w. hao , f. happacher,9 k. hara , a. d. hardman , r. m. harris,7 f. hartmann , j. hauser,4 e. hayashi , j. heinrich , a. heiss , b. hinrichsen , k. d. hoffman , c. holck , r. hollebeek , l. holloway , z. huang , b. t. huffman , r. hughes , j. huston , j. huth , h. ikeda , m. incagli , j. incandela,7 g. introzzi , j. iwai , y. iwata , e. james , h. jensen,7 u. joshi,7 e. kajfasz , h. kambara , t. kamon , t. kaneko , k. karr , h. kasha , y. kato , t. a. keaffaber , k. kelley , m. kelly , r. d. kennedy,7 r. kephart,7 d. kestenbaum , d. khazins,6 t. kikuchi , m. kirk,3 b. j. kim , h. s. kim , s. h. kim , y. k. kim , l. kirsch,3 s. klimenko,8 d. knoblauch , p. koehn , a. kngeter , k. kondo , j. konigsberg,8 k. kordas , a. korytov,8 e. kovacs,1 w. kowald,6 j. kroll , m. kruse , s. e. kuhlmann,1 e. kuns , k. kurino , t. kuwabara , a. t. laasanen , s. lami , s. lammel,7 j. i. lamoureux,3 m. lancaster , m. lanzoni , g. latino , t. lecompte,1 a. m. lee iv,6 s. leone , j. d. lewis,7 m. lindgren,4 t. m. liss , j. b. liu , y. c. liu , n. lockyer , o. long , m. loreti , d. lucchesi , p. lukens,7 s. lusin , j. lys , k. maeshima,7 p. maksimovic , m. mangano , m. mariotti , j. p. marriner,7 g. martignon , a. martin , j. a. j. matthews , p. mazzanti,2 k. mcfarland , p. mcintyre , p. melese , m. menguzzato , a. menzione , e. meschi , s. metzler , c. miao , t. miao,7 g. michail , r. miller , h. minato , s. miscetti,9 m. mishina , s. miyashita , n. moggi , e. moore , y. morita , a. mukherjee,7 t. muller , a. munar , p. murat , s. murgia , m. musy , h. nakada , t. nakaya,5 i. nakano , c. nelson,7 d. neuberger , c. newman - holmes,7 c .- y .
p. ngan , h. niu,3 l. nodulman,1 a. nomerotski,8 s. h. oh,6 t. ohmoto , t. ohsugi , r. oishi , m. okabe , t. okusawa , j. olsen , c. pagliarone , r. paoletti , v. papadimitriou , s. p. pappas , n. parashar , a. parri,9 d. partos,3 j. patrick,7 g. pauletta , m. paulini , a. perazzo , l. pescara , m. d. peters , t. j. phillips,6 g. piacentino , m. pillai , k. t. pitts,7 r. plunkett,7 a. pompos , l. pondrom , j. proudfoot,1 f. ptohos , g. punzi , k. ragan , d. reher , a. ribon , f. rimondi,2 l. ristori , w. j. robertson,6 a. robinson , t. rodrigo , s. rolli , l. rosenson , r. roser,7 t. saab , w. k. sakumoto , d. saltzberg,4 a. sansoni,9 l. santi , h. sato , p. schlabach,7 e. e. schmidt,7 m. p. schmidt , a. scott,4 a. scribano , s. segler,7 s. seidel , y. seiya , f. semeria,2 t. shah , m. d. shapiro , n. m. shaw , p. f. shepard , t. shibayama , m. shimojima , m. shochet,5 j. siegrist , a. sill , p. sinervo , p. singh , k. sliwa , c. smith , f. d. snider,7 j. spalding,7 t. speer , p. sphicas , f. spinella , m. spiropulu , l. spiegel,7 l. stanco , j. steele , a. stefanini , r. strhmer , j. strologas , f. strumia , d. stuart,7 k. sumorok , j. suzuki , t. suzuki , t. takahashi , t. takano , r. takashima , k. takikawa , m. tanaka , b. tannenbaum,4 f. tartarelli , w. taylor , m. tecchio , p. k. teng , y. teramoto , k. terashi , s. tether , d. theriot,7 t. l. thomas , r. thurman - keup,1 m. timko , p. tipton , a. titov , s. tkaczyk,7 d. toback,5 k. tollefson , a. tollestrup,7 h. toyoda , w. trischuk , j. f. de troconiz , s. truitt , j. tseng , n. turini , t. uchida , f. ukegawa , j. valls , s. c. van den brink , s. vejcik iii,7 g. velev , r. vidal,7 r. vilar , i. vologouev , d. vucinic , r. g. wagner,1 r. l. wagner,7 j. wahl,5 n. b. wallace , a. m. walsh , c. wang,6 c. h. wang , m. j. wang , a. warburton , t. watanabe , t. watts , r. webb , c. wei,6 h. wenzel , w. c. wester iii,7 a. b. wicklund,1 e. wicklund,7 r. wilkinson , h. h. williams , p. wilson,7 b. l. winer , d. winn , d. wolinski , j. wolinski , s. worm , x. wu , j. wyss , a. yagil,7 w. yao , k. yasuoka , g. p. yeh,7 p. yeh , j. yoh,7 c. yosef , t. yoshida , i. yu,7 a. zanetti , f. zetti , and s. zucchelli2 .026 in ( cdf collaboration ) .026 in 1 argonne national laboratory , argonne , illinois 60439 + 2 istituto nazionale di fisica nucleare , university of bologna , i-40127 bologna , italy + 3 brandeis university , waltham , massachusetts 02254 + 4 university of california at los angeles , los angeles , california 90024 + 5 university of chicago , chicago , illinois 60637 + 6 duke university , durham , north carolina 27708 + 7 fermi national accelerator laboratory , batavia , illinois 60510 + 8 university of florida , gainesville , florida 32611 + 9 laboratori nazionali di frascati , istituto nazionale di fisica nucleare , i-00044 frascati , italy + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + |
with the development of space infrared ( ir ) astronomy , the precise determination of ir extinction becomes urgent in order to recover the intrinsic colors and spectral energy distributions ( seds ) of heavily obscured sources .
there have been various attempts to measure the ir extinction based on the _ infrared space observatory_(_iso _ ) and _ spitzer space telescope _ since @xcite obtained the mid - ir extinction from several hydrogen recombination lines and demonstrated the absence of the model - predicted pronounced minimum around 7@xmath0 .
this was supported by @xcite based on the isogal database @xcite , and by @xcite based on the data from the _ spitzer _ galactic legacy infrared midplane survey extraordinaire ( glimpse ) legacy program @xcite .
all these results roughly agree with the extinction predicted by the standard interstellar grain model for @xmath7 of @xcite and @xcite .
is the total - to - selective extinction ratio , where @xmath9 , the color excess , is the difference between the extinction in @xmath10 and @xmath11 bands .
] however , so far only a few wave bands have been investigated and the sky coverage is also limited .
no consensus has been reached yet regarding the interstellar extinction at @xmath1258@xmath0 .
recent progress in the ir extinction measurements is made toward star - forming regions mainly based on the _ spitzer _ observations .
thanks to the high sensitivity of _ spitzer _ , deep photometry is now possible and objects that suffer severe extinction are now reachable .
@xcite studied five nearby star - forming regions at mid - ir wavelengths ( 3.6@xmath0 , 4.5@xmath0 , 5.8@xmath0 and 8.0@xmath0 from the infrared array camera [ irac ] , , 4.442@xmath0 , 5.675@xmath0 and 7.760@xmath0 , respectively . ] and 24@xmath0 from the multiband imaging photometer [ mips ] ) .
they confirmed a relatively flat extinction curve at @xmath1248@xmath0 .
@xcite studied a star - forming dense cloud core located in the pipe nebula , and found that the ir extinction in the irac bands of that region also agrees with the @xmath13 model curve and indicates a dust size distribution favoring larger sizes .
although both the _ iso _ and _ spitzer _ measurements agree with each other in that the @xmath1258@xmath0 extinction is relatively flat and lacks the model - predicted minimum around 7@xmath0 of @xcite , there do exist differences among various measurements made for different sightlines . in their sample of five sightlines toward star - forming regions
, @xcite found a clear difference between one sightline and the other four sightlines ( see their table 3 ) .
they derived higher @xmath8 ratios and a flatter wavelength dependence than that of @xcite for the same sightline toward @xmath14 in the galactic plane . from @xmath15200 fields observed in the isogal survey , @xcite analyzed the extinction at 7@xmath0 and 15@xmath0 along @xmath12120 directions .
they found marginal variation of the extinction at 7@xmath0 .
it is commonly believed that , with the parameter @xmath16 increasing in denser regions , the variation of the ultraviolet ( uv ) and visual extinction with wavelength becomes flatter than that of the diffuse interstellar medium ( ism ) which is characterized with a lower @xmath16 @xcite , while the near - ir extinction seems to be universal " , with little variation among different sightlines @xcite .
however , @xcite recently argued against such a universal near - ir extinction .
in addition , @xcite argued that the ir - through - uv galactic extinction curves should not be considered as a simple one - parameter family , whether characterized by @xmath16 as suggested by @xcite or any other parameters .
@xcite presented observational evidence for a small but appreciable variation in @xmath16 with galactic longitude .
he suggested that the most likely explanation for this is a variation in the mean size of the dust in the local spiral arm .
however , unfortunately only a few data points were used in that work and therefore no systematic variation of the extinction with galactic longitude was reported .
@xcite obtained the extinction around 7@xmath0 for 129 different sightlines and no clear variation with galactic longitude was found , although the extinction ratio @xmath17}/a_{{\rm k_{s}}}$ ] does appear to exhibit a tendency of decreasing toward the galactic center where @xmath18@xcite .
the glimpse legacy program surveyed the galactic plane , with a large area coverage ( @xmath19 ) and a detection limit of @xmath1215.513.0@xmath20 from 3.6 to 8.0@xmath0 @xcite .
it provides an opportunity to explore the systematic variation of interstellar extinction in the ir with galactic longitude . in this work
, we explore whether the mid - ir extinction varies among sightlines and how it varies in different interstellar environments based on the _
spitzer_/glimpse database . in [ data ] , the glimpse data used in this work is briefly described .
[ method ] presents the method adopted to derive the extinction . in [ tracer ] we discuss the selection of two different types of tracers ( i.e. red giants and red clump giants ) .
[ results ] reports the resulting extinction ratios @xmath8 and the mean extinction from the total 131 glimpse fields .
also discussed in [ results ] are the comparison of the extinction derived here with previous studies performed by @xcite and @xcite , and the longitudinal variation of the extinction ratios @xmath8 as well as its relation with the galactic spiral arms and the distribution of interstellar dust . in 6
we summarize our major conclusions .
the data used in this work is obtained by the glimpse group .
glimpse is a _ spitzer _ legacy program to carry out an ir survey of the inner galactic plane using the irac camera on board the _ spitzer _ space telescope @xcite .
it spans three cycles , glimpse , glimpse - ii and glimpse-3d ( see @xcite ) .
these glimpse programs observed a large part of the galactic disk , including various sightlines along the galactic plane and providing an opportunity to investigate whether the mid - ir extinction varies from one sightline to another .
the glimpse and glimpse - ii enhanced data have now been released .
these products consist of the highly reliable point source catalogs ( glmc ) , the more complete point source archives ( glma ) and mosaic images of the survey areas . in this study
we will use the glmc catalogs since the sources in these catalogs were selected requiring the reliability to exceed @xmath1299.5% @xcite .
moreover , in the enhanced glimpse version 2.0 catalogs and glimpse - ii version 1.0 catalogs , the point sources were band - merged ( cross - identified ) with the 2mass point source catalog ( see @xcite ) .
they provide both magnitudes and fluxes in the four irac bands and three 2mass bands .
based on the original glimpse and glimpse - ii catalog files , we divided the galactic plane into 131 fields . to @xmath21 and @xmath22 to @xmath23 .
meanwhile , because of the insufficiency of sources , the two fields from @xmath24 to @xmath25 and @xmath26 to @xmath27 are incorporated into the fields @xmath28 to @xmath29 and @xmath30 to @xmath31 , respectively . ] in this work , only the sources with a signal - to - noise ratio s / n@xmath325 in all three 2mass bands and four irac bands are taken into account .
more details about sample selection will be described in [ tracer ] .
benefiting from the numerous detections by _ spitzer_/irac , the number of sources with s / n@xmath325 in a single sky field exceeds 10,000 for almost all the glimpse fields .
the only exception is the glimpse observation strategy validation ( osv ) field , with @xmath33 to @xmath34 and @xmath35 , where the number of sources with s / n@xmath325 is 9625 .
the determination of dust extinction is most commonly made by comparing the flux densities of extincted and un - extincted pairs of stars of the same spectral type @xcite .
@xcite used the h recombination lines detected between 2.59@xmath0 to probe the extinction law in this wavelength range of the galactic center based on a comparison of the observed line fluxes with that expected from the standard case b recombination . here
we adopt the `` color - excess '' method to obtain the extinction .
this method calculates the ratio of two color excesses which can be expressed as following @xmath36 where @xmath37 is the magnitude in the band @xmath38 of interest ; @xmath39 is the magnitude in the reference band @xmath40 ( which is usually taken to be the k or @xmath1 band ) ; @xmath41 is magnitude in the comparison band @xmath42 ( which is usually taken to be the j or h band ) .
therefore , the extinction ratio of the @xmath38 band to the @xmath40 reference band is @xmath43 note that @xmath44 is always greater than 1 ( no matter whether the j band or the h band is chosen as the comparison band ) since it is generally true that @xmath45
. therefore @xmath46 increases with the decreasing of @xmath47 .
the `` color - excess '' method is widely applied to photometric data and can probe deeper than the spectrum - pair method .
most of the ir extinction determination studies are performed using this method . in the color - excess method , a group of sources that have the same intrinsic color indices @xmath48 and
@xmath49 are chosen , @xmath47 is simply the slope of the line that linearly fits the observed color indices @xmath50 and @xmath51 @xcite .
this is a statistical method as it makes use of a large number of sources and reduces the risk of depending on any individual objects with large uncertainties in the determination of their intrinsic color indices .
meanwhile , it is essential for the accuracy of the method to have a homogeneous sample [ i.e. with very small scatter in the color indices @xmath48 and @xmath49 ] . from equations ( [ slope ] ) and ( [ ext ] )
, it can be seen that the determination of @xmath55 , the ratio of the @xmath38-band extinction to the extinction of the reference band @xmath40 , requires the knowledge of @xmath56 , the ratio of extinction at the comparison band @xmath42 to that at the reference band @xmath40 . in previous studies , both the @xmath53 band @xcite and the @xmath54 band @xcite
have been used as the comparison band .
the advantage of taking the @xmath53 band as the comparison band is that @xmath57 is more sensitive to the extinction as @xmath57 is about twice @xmath58 for the same value of @xmath59 ( the extinction at the @xmath1 band ) .
the advantage of choosing the @xmath54 band as the comparison band is that there are more red giants detected in the @xmath54 band than in the @xmath53 band .
this would be particularly important when the sample size is small which could influence the statistics .
but for the sky fields studied here , the sample size is not a problem thanks to the sensitivity of _ spitzer_/irac . in this work
, we therefore adopt the @xmath53 band as the comparison band .
the @xmath53 band was also selected as the comparison band in our previous studies of the isogal fields @xcite .
but in order to compare with the results of @xcite , we also calculate @xmath55 with the @xmath54 band taken as the comparison band .
we take @xmath60 ( with @xmath53 as the comparison band ) or @xmath61 ( with @xmath54 as the comparison band ) as derived by @xcite for sightlines toward the gc . and
@xmath62 for the @xmath63 and @xmath64 sightlines in the galactic plane which are very close to that of @xcite . however , @xcite derived @xmath65 and @xmath66 for the sightlines toward the gc based on red clump stars .
they found a steep power - law for the near - ir extinction ( @xmath67 ) while in literature it is often thought that the near - ir extinction is a `` universal '' power - law @xmath68 with @xmath69 @xcite . ]
in the ir , red giants are appropriate tracers of interstellar extinction and 15@xmath0 by @xcite .
the results were in close agreement with that from the hydrogen recombination lines @xcite . ] for the following reasons : ( 1 ) they have a narrow range of effective temperatures so that the scatter of the intrinsic color indices is small .
the @xmath70 color index , often chosen to describe @xmath71 , has a scatter of only @xmath720.1mag around the central value of @xmath721.2 ; ( 2 ) they are bright in the ir and remain visible even with large extinction and/or at a great distance.mag and @xmath73mag , the distance can be as large as @xmath74kpc , provided there is no extinction along the line of sight toward to the star . ] on average , their absolute @xmath1 magnitudes are as bright as @xmath75mag , making them readily detectable by 2mass even at a distance to the gc ( @xmath128.5@xmath76 ) .
but we should note that evolved red giants may have a circumstellar dust shell which would cause circumstellar extinction and produce ir emission , affecting our understanding of their intrinsic color indices .
accordingly , the selection of red giants is usually based on the mid - ir colors which are barely affected by interstellar extinction . following @xcite , we adopt the following criteria to select red giants as our samples 1 . @xmath77-[4.5 ] < $ ] 0.6 and @xmath78-[8.0 ] < $ ] 0.2 , also adopted by @xcite , we confine ourselves to colors bluer than certain values to exclude the sources with ir excess such as pre - main sequence stars and asymptotic giant branch ( agb ) stars .
this criterion also safely excludes young stellar objects @xcite .
however , it is more complicated for evolved stars . by analyzing the synthetic colors of agb stars obtained by convolving the _ iso _ short wavelength spectrometer ( isosws ) spectra with the irac transmission profiles
, @xcite found that the irac colors of agb stars are similar to that of red giant branch ( rgb ) stars .
therefore the selected samples may be contaminated by some agb stars .
@xcite also found @xmath77-[4.5 ] > 0.3 $ ] for agb stars with a significant mass loss .
the @xmath78-[8.0]$ ] color index is slightly more selective , with young agb stars being redder by @xmath120.2 than most red giants .
nevertheless , the criteria are kept with significant confidence as red giants are much more numerous than agb stars in these relatively blue colors .
this is later proved to be correct in the appearance of the color - color diagram of the sources .
but it should be kept in mind that some agb stars that suffer circumstellar extinction may be included.yrs at rgb and @xmath79yrs at agb @xcite .
the number ratio of rgb / agb would be about two orders of magnitude .
in addition , their color index @xmath78-[8.0]$ ] of agb stars ranges from about 0.0 to 1.0 @xcite , mostly much redder than our critical value 0.2 . combining these two factors , it is a reasonable estimation that the fraction of agb stars in the selected sample should be less than 1 percent . ]
@xmath801.2 or @xmath810.3 .
this criterion excludes foreground dwarf stars since even late - m dwarfs have @xmath820.6 .
both theoretical and observational studies suggest that rgb stars have an intrinsic color of @xmath83 with a dispersion of @xmath120.1mag @xcite . for an m5 giant ,
the calculated @xmath70 is @xmath121.2mag , and @xmath84 is @xmath120.3mag .
band is chosen as the comparison band , this criterion ( @xmath810.3 ) is comparable to that of @xcite , i.e. @xmath810.2 . ] _ in the following , the samples selected in terms of @xmath801.2 are denoted by rg1 , while the ones selected from @xmath810.3 are referred as rg2 . _
good photometric quality .
this criterion guarantees the reliability of the calculated color indices without reducing the number of sources too much so as to degrade the statistical accuracy .
specifically , the photometry quality flags in the 2mass @xmath85 bands are `` aaa '' ( i.e. with s / n@xmath325 ) ; the quality flags in the irac bands are comparable ( in the glimpse catalogs the photometric uncertainty is typically @xmath860.2mag ) .
for all the 131 sky fields studied here , s
/ n@xmath325 is required in all three 2mass bands and four irac bands .
4 . a deviation of @xmath87 from the line fitted to the observed colors @xmath88 ( @xmath89 is the irac band wavelength ) vs. @xmath70 . we used the idl robust fitting program to iterate the fit , rejecting the sources with a deviation larger than 3@xmath90 .
those sources ( with a deviation larger than 3@xmath90 ) were rejected because , from a statistical point of view , it is unlikely for them to follow the linear relation .
furthermore , most of them have relatively large @xmath88 values , indicating the possible presence of extinction arising from circumstellar envelopes @xcite .
@xcite also rejected high-@xmath90 points to exclude some of the extreme red excess sources .
figure[fig1 ] shows the near - ir color - magnitude diagrams ( cmds ) of two typical fields , @xmath91 to @xmath92 and @xmath93 to @xmath94 . in the left panels ( a , c ) ,
the rg1 samples ( based on @xmath801.2 and the other criteria described above ) are denoted by red dots .
the black background points are the sources in those fields with s / n@xmath321 in all three 2mass bands . because we only adopt the sources with s / n@xmath325 in all 7 bands ( 3 2mass bands and 4 irac bands ) , the number of the selected rg samples is much smaller than the total number of sources in these fields .
@xcite chose red clump giants ( rcgs ) , which have a dispersion of @xmath120.3mag in the absolute @xmath1 magnitude and @xmath120.2mag in the near - ir color index @xcite , as a tracer to derive the extinction from the 2mass @xmath53 band to the irac 8.0@xmath0 band .
rcg stars are the equivalent of the horizontal - branch stars for a metal - rich population .
they have narrow distributions in luminosity and color indices , with a weak dependence on metallicity @xcite .
thus their locations in the near - ir cmds could be more clearly distinguished than that of red giants .
this makes them a more homogeneous sample of tracers to obtain the ir extinction . however , rcg stars are fainter than red giants in the ir because of their relatively blue color index ( usually k - type ) and lower absolute luminosity [ @xmath95mag , see @xcite ; or @xmath96mag , see @xcite ] . the rcg stars detected by 2mass are not as distant as red giants .
therefore , they can not trace the extinction as deep as red giants .
due to the weakness of the ir extinction , the traceable depth of extinction is crucial . as shown in @xcite ,
the maximum @xmath57 traced by rcg stars was about 2.6mag , much smaller than the easily - reached 6mag of red giants .
in addition , the color dispersion of rcg stars , @xmath120.2mag , is a little bit larger than that of red giants , @xmath120.1mag @xcite .
nevertheless , in this work rcg stars are also selected as a tracer in deriving the ir extinction and the results are compared with that from the red giant samples . as k2iii giants , the commonly accepted absolute magnitude for rcg stars in the @xmath1 band
is @xmath97 and their intrinsic ir color indices are almost constant , with @xmath98 @xcite .
these characteristics would place them in a well defined narrow stripe in the near - ir cmds ( @xmath70 vs. @xmath1 ) .
however , the distribution of rcg stars in the cmd ( based on their _ observed _ magnitudes and colors ) is much more scattered .
this is because the observed color indices depend only on interstellar extinction , while the observed @xmath1 magnitudes depend not only on the amount of extinction but also on the distance of the star .
in addition , the extinction is highly uneven and is not simply proportional to the distance .
furthermore , there might be contamination of dwarf stars and agbs at the lower end in the cmds .
only @xmath722.55% of the detected sources are dwarfs , but this fraction rises to @xmath721040% for @xmath99 @xcite . ] thus the selection of the rcg stars appears empirical and eye - dependent . in figure[fig1 ] , the right panels show the rcg stripes lying between two red dashed lines .
the extension of the color index @xmath100 of red giants ( see the left panels of fig.[fig1 ] ) clearly connects to that of red clump stars .
the upper right cmd ( fig.[fig1]b ) has a relatively clear branch of red clump stars , but this is not the case for the bottom one . in comparison with the quite scattered distribution of red giants ,
the disk rcg stars stand out by their configured locations in the cmds for a large portion of the sightlines ( e.g. the field @xmath91 to @xmath92 ) , i.e. a narrow stripe from upper - left to lower - right in the cmds caused by interstellar extinction and the increasing distance from the sun . within the rcg strip , the red clump stars consist of relatively homogenous samples .
however , the homogenity depends very much on the specific field , ( e.g. the field with @xmath93 to @xmath94 and @xmath101 does not present a clear stripe of rcg stars in figure[fig1]d ) .
the interstellar extinction is possibly highly nonuniform with distance on the sightline within @xmath93 to @xmath94 , so that a gap appears in the rcg stripe near @xmath102 and @xmath103 .
in addition , a giant branch can be seen at @xmath104 in the lower cmd in figure[fig1]d , which is probably due to the stars in the bulge as suggested by @xcite . although the 2mass point source catalogue is complete down to @xmath105 @xcite , the faintest @xmath1 magnitude in this work is @xmath1213@xmath20 for almost the entire glimpse fields ( because of the s / n@xmath325 requirement for all the 2mass and irac bands ) .
the faintest @xmath1 magnitude of the selected rg samples can only reach @xmath1213@xmath20 ( see fig.[fig1]a , c ) .
for the rcg stars selected from the rcg stripes requiring s / n@xmath325 in all seven bands , the faintest @xmath1 magnitude can also only reach @xmath1213@xmath20 .
@xcite proposed a method to extract rcg stars based on cmds .
they determined an empirical track of rcg stars by linking the peaks of the histograms of all the horizontal cuts in the cmds . applying a limitation of @xmath10613.0 to exclude k dwarf stars , the sources within a deviation of @xmath860.2 mag in @xmath70 from the determined track
are extracted as the selected rcg stars . @xcite and @xcite adopted similar methods to pick up rcg stars . in this work , we take a similar approach .
first , a rough stripe in the @xmath70 vs. @xmath1 cmd was chosen by eye to encircle the preliminary range of the rcg stars .
the eye - selected range is divided into different horizontal cuts with a step of 0.3mag at @xmath1 , then the histograms of each horizontal cut was fitted with gaussian functions to determine the peak color indices @xmath70 for different horizontal @xmath1 cuts .
second , the peak positions in each horizontal cuts are taken as input to delineate the curve fitted with a 2nd - order polynomial for the central location of the rcg stars .
third , the width of the curve derived from the previous step , which corresponds to the scale of the dispersion of the color index , is determined .
unlike @xcite , we do not treat this width as a constant ( because it results not only from the almost constant scatter of the intrinsic color index , but also from the photometric error which increases with decreasing apparent brightness , i.e. increasing magnitude ) .
the increase of dispersion width with the observed magnitude is noticeable from the cmds as well .
so the width adopted in our work differs for different sightlines based on the definition of the rcg stripes in the cmds .
centered with the fitted red clump track , the width ranges from @xmath120.1 to @xmath120.2mag for @xmath107 and increases to @xmath120.20.3mag for @xmath108 ( see fig.[fig1 ] ) .
finally , one more factor to take into account is the enhanced contamination of dwarf stars at fainter magnitudes .
these dwarfs should be mainly earlier than k - type to be bright enough to be visible at relatively large distances ( because their observed colors are much redder than the intrinsic ones which must result from subjecting to substantial extinction , indicating that they can not be nearby ) .
the near - ir intrinsic colors of these dwarf stars are bluer than that of the rcg stars . compared with the rcg stars in the mid - ir
, these dwarfs would have a larger @xmath47 in equation ( [ slope ] ) and lead to a smaller @xmath46 in equation ( [ ext ] ) since @xmath109 . to suppress this effect , the @xmath1 magnitude is limited to 13.0 in @xcite .
we also place an upper limit on the @xmath1 magnitude in order to reduce the confusion from the k dwarfs-[4.5 ] < $ ] 0.6 and @xmath78-[8.0 ] < $ ] 0.2 ( see [ criteria_rg ] ) . ] ( but this upper limit is again different for different sightlines ) . as mentioned earlier , the @xmath1 magnitude can only reach @xmath1213@xmath20 for most of the glimpse fields because of the s / n@xmath325 requirement for all seven bands ( see [ rcg ] ) . consequently , for most of the sky fields , the typical limit on the @xmath1 magnitude is @xmath1212mag . but this limit is much smaller on the sightlines toward the galactic bulge direction , mostly in the range of @xmath110 , where @xmath111 11.5 is generally taken .
the lower cut of the @xmath1 magnitude results in much smaller @xmath70 and much fewer samples than that from the red giant samples .
furthermore , because of the lower @xmath1 magnitude ( @xmath1@xmath11212 ) , the rcg samples in this work can only be used to probe the interstellar extinction from the diffuse ism near the sun . ) appear to lie at a larger distance @xcite and have a @xmath1 magnitude of @xmath121214 @xcite .
because of the @xmath1 magnitude limitation , the rcg stars finally extracted in this work would mostly lie in the galactic disk . ] according to the following expression , @xmath113 if @xmath114 , the limitation of @xmath115 implies that the most distant stars that could be selected in our samples are at @xmath116kpc .
many works have suggested the existence of a linear bar across the galactic center with a half - length of @xmath124.5kpc and a position angle of @xmath1245@xmath117 @xcite , i.e. the closest distance to the sun is @xmath118 ( assuming the distance to the gc from the sun is @xmath119 ) .
thus even if @xmath120 , we can still hardly reach the bulge stars located in the bar beyond @xmath121kpc . in the region within @xmath110 , where @xmath122 is taken
, the distance limit must be closer than @xmath724kpc .
so the interstellar extinction derived from the rcg samples is mainly from the local diffuse ism and it may differ from that derived from red giants ( see [ variation - l ] ) . for most of the 131 glimpse fields ,
the stripes of the rcg stars in the near - ir cmds are clearly visible , indicating that the above - described selection method for rcg stars is reasonably reliable .
but for the fields in the directions to the galactic bulge , there are so many dwarf stars that the stripes of rcg stars are not easily distinguishable , thus only the nearby rcg stars subject to small extinction which lie at the higher end of the stripes can be extracted .
moreover , in the regions within @xmath123 , there is always a giant branch overlapping with the rcg stripe at the low @xmath1 end ( see the vertical branch near @xmath124 in figure[fig1]d ) .
significant contamination occurs at later - type rgb and agb stars @xcite .
the contamination caused by other types of stars degrades the quality of the results derived from the rcg samples .
the reliability of the results varies from field to field .
[ results ] we have obtained the extinction in each irac band ( relative to that of the @xmath1 band ) averaged over all 131 glimpse fields ( see table[tbl - relext ] ) . with red giants as the extinction tracer and the @xmath53 band as the comparison band , the mean extinction ratios are @xmath3}/a _
{ { { \rm k_{s}}}}\approx0.64\pm0.01 $ ] , @xmath4}/a_{{{\rm k_{s}}}}\approx0.63\pm0.01 $ ] , @xmath5}/a_{{{\rm k_{s}}}}\approx0.52\pm0.01 $ ] and @xmath6}/a_{{{\rm k_{s}}}}\approx0.55\pm0.02 $ ] .
the selection criteria of @xmath125 ( rg1 ) or @xmath126 ( rg2 ) makes little difference since the samples selected from these two criteria are almost the same ( see fig.[fig2 ] and the first two rows of table[tbl - relext ] ) .
here we adopt @xmath127 and @xmath60 ( i.e. we take the power index @xmath128 of the near - ir extinction law @xmath68 to be 1.61 ; see @xcite ) . with the @xmath54 band chosen as the comparison band ,
the extinction ratios @xmath129 ( derived from the rg2 samples , see the third row of table[tbl - relext ] ) systematically decrease by an amount of @xmath120.04 at 3.6@xmath0 , @xmath120.04 at @xmath130 , @xmath120.05 at @xmath131 , and @xmath120.05 at 8.0@xmath0 , respectively .
if we take a larger power index [ e.g. @xmath132 as suggested by @xcite ] , the extinction ratios will become smaller . or @xmath54 ,
is chosen as the comparison band .
but the effect is more significant for the former and therefore there will be little difference between the extinction ratios derived using different comparison bands ( @xmath53 or @xmath54 ) . ]
also shown in table[tbl - relext ] are the extinction ratios @xmath129 derived from the rcg samples ( with the @xmath53 band as the comparison band )
. the extinction ratios of the 3.6@xmath0 , 4.5@xmath0 and 5.8@xmath0 bands derived based on the rcg samples are smaller than that based on red giants , while the extinction ratio of the 8@xmath0 band is approximately equal for both samples .
band is an average result . as far as individual regions are concerned , the difference in this band is significant in the galactic bulge region ( see [ variation ] ) . ] as mentioned earlier , the rcg samples may include some dwarf stars at the faint end . since dwarf stars are intrinsically bluer than rcg stars , they may cause the derived extinction to be underestimated ( see eq.[ext ] ) . on the other hand , the extinction would be somewhat overestimated from the red giant samples because of their circumstellar dust .
but for the 8.0@xmath0 band , the extinction derived from red giants may be underestimated because of the possible presence of silicate emission .
silicate emission feature .
the uncertainties discussed here are not included in the quoted errors in table 1 . ]
both @xcite and @xcite have studied the extinction in the glimpse osv field . to allow a direct comparison with their results , in table[ext_osv ]
we tabulate the irac band extinction ( relative to @xmath133 ) derived here as well as theirs .
the extinction values taken from @xcite are for the off - cloud line of sight and derived from rcg stars .
although the extinction results are similar in the irac [ 3.6 ] and [ 4.5 ] bands , the extinction law derived here from the rcg samples for the osv field is much steeper than that derived by @xcite , with much smaller values in the [ 5.8 ] band .
this is probably due to the different methods adopted in selecting rcg stars , such as the limitation on the @xmath1 magnitude or the criterion on color index ( see [ method_rcg ] ) . with the same criteria ( @xmath77-[4.5 ] <
$ ] 0.6 and @xmath78-[8.0 ] < $ ] 0.2 , rg2 ) and the same comparison band @xmath54 as adopted by @xcite , our results are @xmath720.3 larger in the [ 4.5 ] , [ 5.8 ] and [ 8.0 ] bands ( see the third row of table[ext_osv ] ) . the reason for this systematic difference could be attributed to the different power index @xmath128 used as discussed above .
nevertheless , the major reason why the extinction derived here ( the first and second rows in table[ext_osv ] ) is systematically larger than that derived by @xcite for the osv field lies in the choice of the comparison band : in this work we use the @xmath53 band as the comparison band , while @xcite used the @xmath54 band . in view of the different characteristics of rg and rcg stars
, we averaged over both the rg1 and rcg results in order to get a `` mean '' extinction law more or less unrelated to the distance .
the `` mean ''
interstellar extinction in the four irac bands are @xmath134 , @xmath135 , @xmath136 and @xmath137 for @xmath3}/a_{{{\rm k_{s}}}}$ ] , @xmath4}/a_{{{\rm k_{s}}}}$ ] , @xmath5}/a_{{{\rm k_{s}}}}$ ] and @xmath6}/a_{{{\rm k_{s}}}}$ ] , respectively ( see table[tbl - relext ] ) . in figure[fig3 ] we plot our results together with previous observational determinations @xcite and the model extinction calculated for @xmath138 and @xmath7 by @xcite .
it is seen in figure[fig3 ] that our results are in close agreement with previous studies and confirm that the extinction law at @xmath1238@xmath0 is almost flat and lacks the minimum around 7@xmath0 predicted from the silicate - graphite interstellar grain model for @xmath138 , but is close to ( although systematically slightly higher than ) the @xmath7 model curve .
the interstellar extinction law is long known to vary along different sightlines in the optical and uv wavelength range , while the extinction in the near ir was often thought to be `` universal '' ( but this was recently questioned by @xcite ) . from the isogal surveyed fields
, @xcite found marginal variation in @xmath139 .
due to the relatively large uncertainty , the regional variation was not clear . to examine whether there exist regional variations of the wavelength dependence of the ir extinction law , we show in table[131tab ] the extinction ratios @xmath129 in all four irac bands for all 131 glimpse fields , including results from both the rg1 and rcg samples . because of more sample stars , much larger @xmath57 and no limitation on the @xmath1 magnitude , the typical error for the rg1 results is just @xmath120.001 , much smaller than that of the rcg results which is @xmath720.01 .
figure[fig4 ] plots the histograms of the extinction ratios in all 131 glimpse fields and clearly shows that there are a range of extinction ratios for both the rg1 and rcg results .
the range for the rcg results is clearly larger than that for the rg1 results . even for the extinction ratios derived from red giants , the range of variations for any of the irac bands are large enough to exceed the average error of @xmath120.001 , with @xmath120.07 , 0.16 , 0.14 and 0.12 ( see table[tbl - extrange ] ) at 3.6@xmath0 , 4.5@xmath0 , 5.8@xmath0 and 8.0@xmath0 , respectively
this demonstrates that the extinction ( relative to @xmath59 ) in the irac bands varies among different sightlines .
a similar conclusion has already been drawn by @xcite , @xcite , @xcite , and @xcite . the amplitude of variation of @xmath129 is relatively small for the 3.6@xmath0 band and is thus difficult to detect without high quality data .
this may explain why the extinction law in the near - ir was thought to be `` universal '' @xcite .
the variations in the 4.5@xmath0 and 5.8@xmath0 bands should be attributed to continuum extinction , while the variation of the 8@xmath0 extinction largely comes from the variation of the 9.7@xmath0 silicate absorption feature strength ( see fig.[fig3 ] ) among different sightlines .
the fact that the 8@xmath0 dispersion is essentially no greater than that of the 4.5@xmath0 and 5.8@xmath0 bands suggests that the 9.7@xmath0 silicate absorption feature probably does not vary much on these sightlines . as shown in figure[fig5 ] , for the rg1 samples , @xmath4}/a_{{{\rm k_{s}}}}$ ] , @xmath5}/a_{{{\rm k_{s}}}}$ ] and @xmath6}/a_{{{\rm k_{s}}}}$ ]
are all closely correlated with @xmath3}/a_{{{\rm k_{s}}}}$ ] , while the correlation between @xmath6}/a_{{{\rm k_{s}}}}$ ] and @xmath3}/a_{{{\rm k_{s}}}}$ ] for the rcg samples is not as tight as that of the [ 4.5 ] and [ 5.8 ] bands with the [ 3.6 ] band .
this is probably due to the contamination of o - rich agb stars at the lower end of rcg stripes : the extinction in the 3.6@xmath0 , 4.5@xmath0 , and 5.8@xmath0 bands largely comes from the dust which produces the @xmath1216@xmath0 continuum extinction ( e.g. graphite , see fig.8 of @xcite ) , in contrast , the 8.0@xmath0 extinction arises mainly from the 9.7@xmath0 silicate absorption feature .
if the 3.6@xmath0 , 4.5@xmath0 , and 5.8@xmath0 extinction is caused by interstellar dust while the 8@xmath0 extinction is partly from interstellar dust and partly from o - rich agb stars , one would expect that @xmath6}/a_{{{\rm k_{s}}}}$ ] does not correlate with @xmath3}/a_{{{\rm k_{s}}}}$ ] as closely as @xmath4}/a_{{{\rm k_{s}}}}$ ] and @xmath5}/a_{{{\rm k_{s}}}}$ ] correlate with @xmath3}/a_{{{\rm k_{s}}}}$ ] .
we note that the absorption features of polycyclic aromatic hydrocarbon ( pah ) at 6.2 , 7.7 , and 8.6@xmath0 are too weak to contribute to the discontinuities between rcg and rgs ( see fig.16 and 11 of @xcite , and 10.3 of @xcite . unlike previous studies of the isogal fields in which no clear structural distribution along galactic longitude or latitude was found for the extinction at 7@xmath0 and 15@xmath0 @xcite , in this work we see an interesting distribution of extinction .
whichever extinction tracer [ red giants ( rg1 , rg2 ) or rcg stars ] or sample - selection - criterion is used , we always see an uneven distribution of extinction ratios in each irac band with galactic longitude ( see fig.[fig6 ] ) .
since the results from the rg1 and rg2 samples are approximately the same , we only show that of the rg1 ( and rcg ) samples in figure[fig6 ] .
[ [ longitudinal - extinction - profiles - derived - from - rg1-or - rcg - samples ] ] longitudinal extinction profiles derived from rg1 or rcg samples : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + although the extinction values derived from the rcg stars are systematically smaller than that from the red giants as discussed in the previous section , the structural features appear at almost the same positions outside the galactic bulge region for both samples .
the common features include , dips around @xmath140 , @xmath141 , @xmath142 , @xmath143 , peaks around @xmath144 ( see fig .
[ fig6 ] ) .
however , in the bulge direction within @xmath110 , the two samples yield different variation patterns .
this can be understood in view of the fact that the rcg samples selected in this work do not trace as deeply as the red giant samples to the galactic bulge due to the limiting magnitude of @xmath145 on the rcg stars in these regions . as mentioned earlier
, the selected rcg samples can only reach a distance of @xmath124kpc from us to the bulge direction even if there is no extinction .
in contrast , the red giants trace the structure of the bulge such as the bar and the distant spiral arms , which shows some of the differences in the longitudinal distribution of @xmath129 toward the gc .
another difference between the results from the rg1 and rcg samples lies in the regions between @xmath146 .
similarly , the rcg samples can only probe the interstellar extinction of the ism of nearby spiral arms ( e.g. the sagittarius - carina arm located at @xmath122kpc from the sun ) , but not the extinction from the more distant @xmath123kpc - molecular ring between @xmath146 which contains a significant amount of co and dust @xcite .
ring is at @xmath147 .
@xcite suggested the northern tangent is near @xmath148 .
@xcite adopted tangent directions of @xmath149 for the 3@xmath76 arm based on their composite co survey . with @xmath150 ,
the near arm is at a distance of @xmath125.2@xmath76 @xcite .
] meanwhile , the rcg stars can only trace the crux - scutum arm at @xmath151 @xcite , not the more distant norma arm @xcite , thus the longitudinal profiles from rcgs and rgs are also different along this direction .
[ [ relationship - with - the - galactic - spiral - arms ] ] relationship with the galactic spiral arms : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + because the dust properties of the spiral arm regions differ from that of the inter - arm regions @xcite , the variation of extinction with galactic longitude could be related to the spiral structure .
we label the tangent positions of the spiral arms in figure[fig6 ] with long black arrows .
dust emission supports the four - arm model @xcite .
note that both the extinction and 240@xmath0 emission are from dust .
@xcite evaluated , compiled and compared previous works on the determination of the position parameters of the spiral arms . in his four - arm structure model , @xcite placed the mean longitudes of tangent from the sun to the carina - sagittarius arm , the crux - scutum arm , the norma arm , the start of the perseus arm , the scutum - crux arm , and the sagittarius - carina arm at @xmath152 and @xmath153 , respectively ( see his table2 and fig.2 ) . ]
it is seen that the locations of the spiral arms coincide with the dips of the extinction ratios @xmath129 derived from the rg1 samples . from the rg1 samples in the following discussion . ] in particular , the coincidence of the locations of the @xmath129 minimum and the spiral arms is outstanding at negative longitudes ( e.g. the crux - scutum arm at @xmath140 , the norma arm at @xmath154 , and the southern tangent of the 3@xmath76 ring around @xmath155 ) . for the positive longitudes ,
the locations of the scutum - crux arm and the sagittarius - carina arm also coincide with the valleys of @xmath129 .
for the broad dip around @xmath142 , there is no clear tangent direction to any arms , but this direction points to the start of the norma arm and the scutum - crux arm ( see fig.2 of @xcite ) , and probably one end of the galactic bar . ) and orientation ( @xmath156 ) of the galactic bar from the m and k giants in the glimpse survey . ]
@xcite found an extra dip around @xmath142 in the longitude distribution of the 2.2@xmath0 flux obtained by the diffuse infrared background experiment ( dirbe ) on board the _ cosmic background explorer _ ( _ cobe _ ) ( see their fig.1 ) .
they speculated that it results from a large dust cloud in a spiral arm or a molecular ring .
in addition , they also found a broad 2.2@xmath0 valley between about @xmath157 and @xmath158 , which corresponds to the expected position of the 3-kpc ring .
again , the dip of @xmath129 around @xmath159 may be consistent with the location of the 3@xmath76 ring .
the dip of the longitudinal distribution of @xmath129 at @xmath160 appears shifted from the position of the norma arm at @xmath161 .
but we note that the longitudinal position of the tangent to the norma arm is not precisely known and the reported positions range from @xmath162 to @xmath163 [ @xcite ; e.g. @xmath164 @xcite , @xmath165 @xcite , @xmath166 @xcite ] .
this probably also explains why the dip of the longitudinal distribution of @xmath129 at @xmath167 does not precisely match the tangent direction to the sagittarius - carina arm at `` @xmath168 '' ( for which the reported positions range from @xmath169 to @xmath170 , @xcite ) .
why does @xmath129 reach its minimum in the galactic spiral arm regions ?
a simple explanation would be grain growth : in spiral galaxies , interstellar gas and dust are concentrated in the inner edges of spiral arms ( probably caused by the spiral density wave ; @xcite ) .
@xcite argued that as the interarm matter encounters the density wave potential minimum , the gas is compressed by the density wave shock ( with a speed of a few tens kms@xmath171 ; @xcite ) at the inner edge of the spiral arm .
this would lead not only to a larger number density of gas and dust , but also an increase of grain size by accretion and coagulation . as shown in figure7 ,
if the dust in the spiral arm regions ( where giant molecular clouds are strongly concentrated ) grows to @xmath1720.20.3@xmath0 , the wavelength dependence of @xmath129 in the wavelength range of 28@xmath0 becomes steeper ( i.e. smaller ) than that of dust of radii @xmath173 ( mean size for the dust in the diffuse ism ) for amorphous silicate , amorphous carbon , and graphite .
this explains why the @xmath129 values are small in the galactic spiral arms for all four irac bands .
[ [ relationship - with - the - dust - ir - emission ] ] relationship with the dust ir emission : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in figure[fig8 ] we plot the longitudinal distribution of the galactic 240@xmath0 emission from the dirbe galactic plane maps with @xmath3}/a_{{\rm k_{s}}}$ ] , as well as the locations of the spiral arms @xcite .
we see that most of the dips of the @xmath3}/a_{{\rm k_{s}}}$ ] profile coincide with the peaks of the 240@xmath0 emission ( e.g. the features around @xmath174 ) .
@xcite found that the features in the 240@xmath0 emission profile from dust can be identified with spiral arm tangents .
similarly , most of the dips of the extinction profile can be understood in terms of the locations of the spiral arms ( e.g. the 240@xmath0 emission peak and the @xmath3}/a_{{\rm k_{s}}}$ ] dip at @xmath175 vs. the crux - scutum arm , the @xmath165 feature vs. the norma arm , the @xmath176 feature vs. the scutum - crux arm , the @xmath177 feature vs. the sagittarius - carina arm ; the features around @xmath178 and @xmath159 vs. the 3@xmath76 ring or arm ) : in the spiral arm regions which are rich in star - forming giant molecular clouds and hii regions , not only is the dust size relatively larger because of accretion and coagulational growth ( which explains the coincidence of the dips of @xmath129 with the locations of the spiral arms ) , but the dust number density is also higher because of the gas and dust concentration in the inner edges of spiral arms caused by the spiral density wave , and the starlight intensity which illuminates the dust causing it to emit at 240@xmath0 is also higher because of the star formation activity occurring in the arm . both the increased dust concentration and starlight intensity would result in a higher 240@xmath0 emission .
[ [ the - galactic - bulge - region ] ] the galactic bulge region : + + + + + + + + + + + + + + + + + + + + + + + + + + in all four irac bands , the extinction ratios @xmath129 derived from red giants decrease for the sightlines toward the bulge region and reaches their minimum values near @xmath179 , while it is the opposite for that derived from the rcg samples .
this difference is probably due to the limited distance which the rcg samples can probe , as discussed earlier in 4.2.2 .
in addition , the extinction is not simply a linear function of distance because of the nonuniform distribution ( i.e. grain density , size , and composition ) of interstellar dust along these directions . ) to the 9.7@xmath0 si o optical depth ( @xmath180 ) and the ratio of @xmath181 to the 3.4@xmath0 c h optical depth ( @xmath182 ) show considerable variations from the local diffuse ism to the gc ( see @xcite ) .
@xcite explained this by assuming that the abundance of the c h carrier ( relative to other dust components ) gradually increases from the local ism toward the gc . using the rcg stars in the gc within @xmath18 as a probe
, @xcite found that the interstellar extinction of these areas is indeed highly nonuniform .
also the conventional linear relation between extinction and distance of @xmath183 was obtained for stars within @xmath121000pc of the sun and within @xmath12100pc of the galactic plane @xcite . toward the galactic bulge , in view of the large scale structure of the ism ,
the small - scale patchy distribution of dust seems well established ( see @xcite ) . ]
thus the extinction probed by red giants and rcg stars in this work may be very different for the sightlines toward the bulge region .
the decreasing trend of @xmath129 toward the gc suggests that the extinction law toward the gc may be different from that of the local diffuse ism or star - forming regions .
recently , using the bulge rcg stars , @xcite obtained the @xmath121.28@xmath0 interstellar extinction law toward the gc .
they found an appreciably steeper extinction law compared with that of the off - cloud regions in the galactic plane of @xcite and the star - forming regions of @xcite .
@xcite studied the isogal fields using red giants and found no clear structural distribution of the 7@xmath0 extinction along galactic longitude .
we re - analyzed the extinction around 7@xmath0 by smoothing the distribution through a 3-degree bin .
the results ( based on the iso / sws 6.7@xmath0 lw2 , 6.8@xmath0 lw5 , and 7.7 @xmath0 lw6 data ) are shown in figure[fig9 ] .
the smoothed , re - analyzed data appear to display a longitude variation of @xmath184}/a_{{\rm k_{s}}}$ ] , with the local minimum at the gc as the most notable feature . in particular , the extinction ratio @xmath185 shows a profile similar to that derived from the rg1 samples .
using red giants and red clump giants as tracers , we have derived @xmath129 , the extinction ( relative to the 2mass @xmath1 band ) in the four irac bands , [ 3.6 ] , [ 4.5 ] , [ 5.8 ] and [ 8.0]@xmath0 for 131 glipmse fields along the galactic plane within @xmath2 , based on the data obtained from the _ spitzer_/glipmse legacy program and the 2mass survey project .
the principal results of this paper are the following : 1 . the mean extinction in the irac bands ( normalized to the 2mass @xmath1 band ) , @xmath3}/a_{{\rm k_{s}}}\approx0.63\pm0.01 $ ]
, @xmath4}/a_{{\rm k_{s}}}\approx0.57\pm0.03 $ ] , @xmath5}/a_{{\rm k_{s}}}\approx0.49\pm0.01 $ ] , and @xmath6}/a_{{\rm k_{s}}}\approx0.55\pm0.03 $ ] , exhibits little variation with wavelength and lacks the minimum at @xmath127@xmath0 predicted from the standard interstellar grain model for @xmath138 .
this is consistent within errors with previous observational determinations based on _ iso _ and _ spitzer _ data and with that predicted from the grain model for @xmath7 of @xcite .
2 . the wavelength dependence of interstellar extinction in the mid - ir varies from one sightline to another , suggesting that there may not exist a `` universal '' ir extinction law .
3 . there exist systematic variations of extinction with galactic longitude which appears to correlate with the locations of spiral arms and with the variation of the 240@xmath0 dust emission .
this can be understood in terms of larger grain sizes ( arising from coagulational growth ) , enhanced dust concentration , and higher starlight intensities in the spiral arm regions .
alves , d. r. 2000 , , 539 , 732 allen , l. e. , et al.2004 , , 154 , 363 benjamin , r. a. , et al .
2003 , , 115 , 953 benjamin , r. a. , et al .
2005 , , 630 , l149 benjamin , r. a. 2008 , in asp conf . ser . 387 , massive star formation : observations confront theory , ed .
h. beuther , h. linz , & th .
henning ( san francisco , ca : asp ) , 375 bertelli , g. , bressan , a. , chiosi , c. , fagotto , f. , & nasi , e. 1994 , , 106 , 275 bloemen , j. b. , deul , e. r. , & thaddeus , p. 1990
, , 233 , 437 bronfman , l. 2008 , , 313 , 81 cabrera - lavers , a. , hammersley , p. l. , gonzlez - fernndez , c. , lpez - corredoira , m. , garzn , f. , & mahoney , t. j. 2007 , , 465 , 825 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 chapman , n. l. , mundy , l. g. , lai , s .-
p . , & evans , n. j. 2009 , , 690 , 496 churchwell , e. , et al .
2006 , glimpse quality assurance v1.0 .
churchwell , e. , et al .
2009 , , 121 , 213 cohen , r. s. , cong , h. , dame , t. m. , thaddeus , p. 1980 , , 239 , l53 cutri , r. m. , et al.2003 , the irsa 2mass all - sky point source catalog , nasa / ipac infrared science archive . http://irsa.ipac.caltech.edu/applications/gator/ dame , t. m. , hartmann , d. , & thaddeus , p. 2001
, , 547 , 792 dame , t. m. , & thaddeus , p. 2008
, , 683 , l143 demers , s. , irwin , m. j. , & gambu , i. 1994 , mnras , 266 , 7 draine , b. t. , & lee , h. m.1984 , , 285 , 89 draine , b. t. 1989 , in infrared spectroscopy in astronomy , ed .
kaldeich ( paris , esa ) , 93 draine , b. t. 2003 , , 41 , 241 draine , b. t. , & li , a. 2007 , , 657 , 810 drimmel , r. 2000 , , 358 , l13 drimmel , r. , cabrera - lavers , a. , & lpez - corredoira , m. 2003 , , 409 , 205 fitzpatrick , e. l. 2004 , in asp conf . ser .
309 , astrophysics of dust , ed .
a. n. witt , g. c. clayton , & b. t. draine ( san francisco , ca : asp ) , 33 fitzpatrick , e. l. , & massa , d. 2009 , , 699 , 1209 flaherty , k. , pipher , j. , megeath , s. , wisnton , e. , gutermuth , r. , muzerolle , j. , allen , l. , & fazio , g. 2007 , , 663 , 1069 gao , j. , jiang , b. w. , & li , a. 2009 , earth , planets & space , in press glass , i. , ganesh , s. , alard , c. , blommaert , j. , gilmore , g. , lloyd evans , t. , omont , a. , schultheis , m. , & simon , g. 1999 , , 308 , 127 greenberg , j. m. 1970 , in iau symp .
39 , interstellar gas dynamics , ed .
habing ( dordrecht : reidel ) , 306 greenberg , j. m. , & li , a. 1995 , in nato asi conf .
469 , the opacity of spiral disks , ed .
davies & d. burstein ( dordrecht : kluwer ) , 19 groenewegen , m. 2006 , , 448 , 181 hammersley , p. l. , garzn , f. , mahoney , t. j. , and calbet , x. 1994 , , 269 , 753 hammersley , p. l. , garzn , f. , mahoney , t. j. , lpez - corredoira , m. , & torres , m. a. p. 2000 , , 317 ,
l45 indebetouw , r. , et al .
2005 , , 619 , 931 jiang , b.w . ,
omont , a. , ganesh , s. , simon , g. , & schuller , f. 2003 , , 400 , 903 jiang , b.w . , gao , j. , omont , a. , schuller , f. , & simon , g. 2006 , , 446 , 551 li , a. , & draine , b. t. 2001 , , 554 , 778 lpez - corredoira , m. , cabrera - lavers , a. , garzn , f. , & hammersley , p.l .
2002 , , 394 , 883 lutz , d. , et al .
1996 , , 315 , l269 lutz , d. 1999 , in the universe as seen by iso , ed .
p. cox & m. f. kessler ( esa sp-427 ; noordwijk : esa ) , 623 marengo , m. , hora , j. l. , barmby , p. , willner , s. p. , allen , l. e. , schuster , m. t. , & fazio , g. g. 2007 , why galaxies care about agb stars : their importance as actors and probes , 378 , 80 marengo , m. , reiter , m. , & fazio , g. g. 2008 , evolution and nucleosynthesis in agb stars , 1001 , 331 mcclure , m. 2009 , , 693 , l81 megeath , s. t. , gutermuth , r. a. , allen , l. e. , pipher , j. l. , myers , p.c . &
fazio , g. g. 2004 , , 154 , 367 nishiyama , s. , et al .
2006 , , 638 , 839 nishiyama , s. , et al .
2006 , , 647 , 1093 nishiyama , s. , tamura , m. , hatano , h. , kato , d. , tanab , t. , sugitani , k. , & nagata , t. 2009 , , 696 , 1407 omont , a. , et al .
1999 , , 348 , 755 rieke , g. , & lebofsky , m. 1985 , , 288 , 618 roberts , w. w. 1969 , , 158 , 123 roberts , w. w. , roberts , m. s. , & shu , f. h. 1975 , , 196 , 381 roberts , w. w. , jr . , huntley , j. m. , & van albada , g. d. 1979 , , 233 , 67 romn - ziga , c. g. , lada , c. j. , muench , a. , & alves , j. f. 2007 , , 664 , 357 sandford , s. a. , pendleton , y. j. , & allamandola , l. j. 1995 , , 440 , 697 spitzer , l. 1978 , physical processes in the interstellar medium ( new york : wiley - interscience ) , 46 valle , j. p. 2008
, , 135 , 1301 van loon , j. t. , et al .
2003 , , 338 , 857 van woerden , h. , rougoor , g. w. , & oort , j. h. 1957 , academie des sciences paris comptes rendus , 244 , 1691 vassiliadis , e. , & wood , p. r. 1993 , , 413 , 641 wainscoat , r. , cohen , m. , volk , k. , walker , h. , & schwartz , d. 1992 , , 83 , 146 weingartner , j.c . , & draine , b.t .
2001 , , 548 , 296 whittet , d. c. b. 1977 , , 180 , 29 @xmath92 and @xmath186 ( upper panels : a , b ) , and @xmath93@xmath94 and @xmath35 ( lower panels : c , d ) . denoted by black dots , the sources in these fields with s / n@xmath321 in all three 2mass bands are all plotted . in the left panels ( a , c ) ,
red dots denotes the red giants selected satisfying @[email protected] ( rg1 ) , s / n@xmath325 and the other criteria ( see [ criteria_rg ] ) . in the right panels ( b , d ) , the selected rcg stars lie in between the two red dashed lines , with the center characterized by @xmath187 and @xmath188 for @xmath91@xmath92 and @xmath93@xmath94 , respectively .
the half widths of the rcg stripes are @xmath120.3@xmath20 at @xmath189 for panel b , @xmath120.2@xmath20 at @xmath189 for panel d.,width=480 ] ( solid line ) and for @xmath7 ( dot - dashed line ) of @xcite ( wd01 ) .
the dotted line smoothly connects the mean extinction ratios @xmath129 derived in this work and that of @xcite at 6.7@xmath0 .
the red bars plot the uncertainty ranges of @xmath129 in each irac band .
[ fig3],width=336 ] in the four irac bands obtained from the rg1 ( asterisks ) and rcg samples ( crosses ) .
the typical error for the rg1 results is @xmath120.001 , which is much smaller than @xmath120.01 for the rcg results .
the solid and dashed lines are drawn with the smoothed results .
the solid vertical arrows show the tangent directions of the spiral arms at @xmath190 and @xmath191 @xcite .
the dashed arrow shows the tangent direction of the 3@xmath76 ring at @xmath192 @xcite .
[ fig6],width=480 ] band ) @xmath8 of spherical grains of radii @xmath173 and 0.3@xmath0 of amorphous silicate , graphite , or amorphous carbon composition .
the @xmath8 values in four irac bands are marked with filled circles .
we see that if the dust in the galactic spiral arm regions grows to @xmath193 , one would expect a steeper extinction law ( i.e. smaller @xmath8 ratios ) than that of the diffuse ism for which the mean dust size is @xmath120.1@xmath0 .
[ fig7],width=336 ] } /a_{{\rm
k_{s}}}$ ] and the 240@xmath0 dust emission from the dirbe galactic plane maps ( averaged over the latitude interval @xmath194 ) .
most of the dips on the @xmath3}/a_{{\rm k_{s}}}$ ] profile coincide with the 240@xmath0 emission peak .
[ fig8],width=336 ] } /a_{{\rm k_{s}}}$ ] ( solid line ) and @xmath17}/a_{{\rm k_{s}}}$ ] ( dashed line ) .
the latter was obtained from smoothing , re - analyzing the iso lw2 ( 6.7@xmath0 ) , lw5 ( 6.8@xmath0 ) and lw6 ( 7.7@xmath0 ) data of the isogal fields @xcite .
[ fig9],width=336 ] l|cccc tracers & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a _ { { \rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] + red giants ( rg1 , j ) & [email protected] & [email protected] & [email protected] & [email protected] + _ red giants ( rg2 , j ) _ & _ [email protected] _ & _ [email protected] _ & _ [email protected] _ & _ [email protected] _ + _ red giants ( rg2 , h ) _ & _ [email protected] _ & _ [email protected] _ & _ [email protected] _ & _ [email protected] _ + red clump giants ( rcg , j ) & [email protected] & [email protected] & [email protected] & [email protected] + mean & [email protected] & [email protected] & [email protected] & [email protected] + nishiyama09 & [email protected] & [email protected] & [email protected] & [email protected] + flaherty07 & [email protected] & [email protected] & [email protected] & [email protected] + indebetouw05 & [email protected] & [email protected] & [email protected] & [email protected] + l|cccc tracers & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a _ { { \rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] + red giants ( rg1 , j ) & [email protected] & [email protected] & [email protected] & [email protected] + red giants ( rg2 , j ) & [email protected] & [email protected] & [email protected] & [email protected] + red giants ( rg2 , h ) & [email protected] & [email protected] & [email protected] & [email protected] + red clump giants ( rcg , j ) & [email protected] & [email protected] & [email protected] & [email protected] + indebetouw05 & [email protected] & [email protected] & [email protected] & [email protected] + flaherty07 & [email protected] & [email protected] & [email protected] & [email protected] + p1.7cm|p1.3cmp1.4cm|p1.5cmp1.5cmp1.5cmp1.6cm|p1.3cmp1.3cmp1.3cmp1.3cmp0.8 cm & & & + field name & @xmath196(deg ) & @xmath197(deg ) & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & max(@xmath1 ) + glmiic_l000 & [ 0,1 ] & [ -2,+2 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l001 & [ 1,2 ] & [ -2,+2 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l002 & [ 2,3 ] & [ -1.5,+1.5 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l003 & [ 3,4 ] & [ -1.5,+1.5 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l004 & [ 4,5 ] & [ -1.5,+1.5 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l005 & [ 5,6 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmiic_l006 & [ 6,7 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmiic_l007 & [ 7,8 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmiic_l008 & [ 8,9 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmxc_l009 & [ 9,10 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmic_l010 & [ 10,11 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmic_l011 & [ 11,12 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmic_l012 & [ 12,13 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l013 & [ 13,14 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l014 & [ 14,15 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l015 & [ 15,16 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l016 & [ 16,17 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l017 & [ 17,18 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l018 & [ 18,19 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l019 & [ 19,20 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l020 & [ 20,21 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l021 & [ 21,22 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l022 & [ 22,23 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l023 & [ 23,24 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l024 & [ 24,25 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l025 & [ 25,26 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l026 & [ 26,27 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l027 & [ 27,28 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l028 & [ 28,29 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l029 & [ 29,30 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l030 & [ 30,31 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l031 & [ 31,32 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l032 & [ 32,33 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l033 & [ 33,34 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 13.0 + glmic_l034 & [ 34,35 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + + & & & + field name & @xmath196(deg ) & @xmath197(deg ) & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & max(@xmath1 ) + glmic_l035 & [ 35,36 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l036 & [ 36,37 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l037 & [ 37,38 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l038 & [ 38,39 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l039 & [ 39,40 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l040 & [ 40,41 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l041 & [ 41,42 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l042 & [ 42,43 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l043 & [ 43,44 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l044 & [ 44,45 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l045 & [ 45,46 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l046 & [ 46,47 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l047 & [ 47,48 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l048 & [ 48,49 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l049 & [ 49,50 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l050 & [ 50,51 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l051 & [ 51,52 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l052 & [ 52,53 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l053 & [ 53,54 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l054 & [ 54,55 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l055 & [ 55,56 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l056 & [ 56,57 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l057 & [ 57,58 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l058 & [ 58,59 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l059 & [ 59,60 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l060 & [ 60,61 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l061 & [ 61,62 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l062 & [ 62,63 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l063 & [ 63,64 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l064 & [ 64,65 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l283 & [ 283.8,284.6 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l295 & [ 295,296 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l296 & [ 296,297 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l297 & [ 297,298 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l298 & [ 298,299 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + + & & & + field name & @xmath196(deg ) & @xmath197(deg ) & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & max(@xmath1 ) + glmic_l299 & [ 299,300 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l300 & [ 300,301 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l301 & [ 301,302 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l302 & [ 302,303 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l303 & [ 303,304 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l304 & [ 304,305 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l305 & [ 305,306 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l306 & [ 306,307 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.5 + glmic_l307 & [ 307,308 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l308 & [ 308,309 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l309 & [ 309,310 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l310 & [ 310,311 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l311 & [ 311,312 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l312 & [ 312,313 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l313 & [ 313,314 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l314 & [ 314,315 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l315 & [ 315,316 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l316 & [ 316,317 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l317 & [ 317,318 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l318 & [ 318,319 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l319 & [ 319,320 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l320 & [ 320,321 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l321 & [ 321,322 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l322 & [ 322,323 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l323 & [ 323,324 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l324 & [ 324,325 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l325 & [ 325,326 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l326 & [ 326,327 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l327 & [ 327,328 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l328 & [ 328,329 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l329 & [ 329,330 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l330 & [ 330,331 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l331 & [ 331,332 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l332 & [ 332,333 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l333 & [ 333,334 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + + & & & + field name & @xmath196(deg ) & @xmath197(deg ) & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & @xmath3}/a_{{\rm k_{s}}}$ ] & @xmath4}/a_{{\rm k_{s}}}$ ] & @xmath5}/a_{{\rm k_{s}}}$ ] & @xmath6}/a_{{\rm k_{s}}}$ ] & max(@xmath1 ) + glmic_l334 & [ 334,335 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l335 & [ 335,336 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l336 & [ 336,337 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l337 & [ 337,338 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l338 & [ 338,339 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l339 & [ 339,340 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l340 & [ 340,341 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l341 & [ 341,342 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l342 & [ 342,343 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l343 & [ 343,344 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l344 & [ 344,345 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l345 & [ 345,346 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l346 & [ 346,347 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l347 & [ 347,348 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l348 & [ 348,349 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmic_l349 & [ 349,350 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmxc_l350 & [ 350,351 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l351 & [ 351,352 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l352 & [ 352,353 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l353 & [ 353,354 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmiic_l354 & [ 354,355 ] & [ -1,+1 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 12.0 + glmiic_l355 & [ 355,356 ] & [ -1.5,+1.5 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l356 & [ 356,357 ] & [ -1.5,+1.5 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l357 & [ 357,358 ] & [ -1.5,+1.5 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l358 & [ 358,359 ] & [ -2,+2 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + glmiic_l359 & [ 359,360 ] & [ -2,+2 ] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] & 11.5 + c|ccc @xmath89 ( @xmath0 ) & minimum & maximum & range + 3.6 ( 3.545 ) & 0.61 & 0.68 & 0.07 + 4.5 ( 4.442 ) & 0.57 & 0.73 & 0.16 + 5.8 ( 5.675 ) & 0.48 & 0.62 & 0.14 + 8.0 ( 7.760 ) & 0.47 & 0.59 & 0.12 + 7.0 & 0.37 & 0.55 & 0.18 + 15.0 & 0.19 & 0.54 & 0.35 + | based on the data obtained from the _ spitzer_/glipmse legacy program and the 2mass project , we derive the extinction in the four irac bands , [ 3.6 ] , [ 4.5 ] , [ 5.8 ] and [ 8.0]@xmath0 , relative to the 2mass @xmath1 band ( at 2.16@xmath0 ) for 131 glipmse fields along the galactic plane within @xmath2 , using red giants and red clump giants as tracers . as a whole , the mean extinction in the irac bands ( normalized to the 2mass @xmath1 band ) , @xmath3}/a_{{\rm k_{s}}}\approx0.63\pm0.01 $ ] , @xmath4}/a_{{\rm k_{s}}}\approx0.57\pm0.03 $ ] , @xmath5}/a_{{\rm k_{s}}}\approx0.49\pm0.03 $ ] , @xmath6}/a_{{\rm k_{s}}}\approx0.55\pm0.03 $ ] , exhibits little variation with wavelength ( i.e. the extinction is somewhat flat or gray ) .
this is consistent with previous studies and agrees with that predicted from the standard interstellar grain model for @xmath7 by @xcite . as far as individual sightline is concerned , however , the wavelength dependence of the mid - infrared interstellar extinction @xmath8 varies from one sightline to another , suggesting that there may not exist a `` universal '' ir extinction law .
we , for the first time , demonstrate the existence of systematic variations of extinction with galactic longitude which appears to correlate with the locations of spiral arms as well as with the variation of the far infrared luminosity of interstellar dust . |
our starting point is benjamin weiss work @xcite where he showed that there exist non - trivial translation - invariant probability measures on the space of entire functions of one complex variable ( a formal definition of such measures will be given several lines below ) .
actually , weiss showed that there is an abundance of such measures .
another approach to the construction of such measures was suggested by tsirelson @xcite .
tsirelson worked in a somewhat different and simpler context .
the works of weiss and tsirelson raise a number of intriguing questions which lie at the crossroads of complex analysis and ergodic theory . here , we address some of these questions .
[ [ section ] ] let @xmath0 denote the space of entire functions with the topology generated by the semi - norms @xmath1 where @xmath2 runs over all compact subsets of @xmath3 , and let @xmath4 be the borel sigma - algebra generated by this topology . then @xmath3 acts on @xmath5 by translations : @xmath6 the probability measure @xmath7 on @xmath5 is called _ translation - invariant _ if it is invariant with respect to this action .
the translation - invariant measure @xmath7 is called _ non - trivial _ if the set of all constant functions in @xmath0 has measure zero ( the constant functions are fixed points of the action @xmath8 ) . due to @xcite , we know that non - trivial translation invariant probability measures on @xmath5 exist . in what follows , we retain the notation @xmath7 for such measures .
after some reflection it becomes plausible that entire functions from the borel support of @xmath7 must grow sufficiently fast and that @xmath7 must have heavy tails .
the goal of this work is to justify these statements .
[ [ section-1 ] ] for an entire function @xmath9 we put @xmath10 where @xmath11 .
[ thm1 ] \(a ) let @xmath7 be a non - trivial translation - invariant probability measure on the space of entire functions .
then , for @xmath7-a.e
. function @xmath9 and for every @xmath12 ,
@xmath13 \(b ) there exists a non - trivial translation - invariant probability measure on the space of entire functions such that , for @xmath7-a.e .
function @xmath9 and for every @xmath12 , @xmath14 [ [ subsub-1.2.1 ] ] the proof of the first part of theorem [ thm1 ] relies on a growth estimate of subharmonic functions , which might be of independent interest . to bring this estimate
we introduce some notation . * till the end of section [ subsub-1.2.1 ] , we assume that _ all squares denoted by @xmath15 and @xmath16 have all four vertices with integer - valued coordinates_. let @xmath17 be a non - negative subharmonic function on a neighbourhood of the square @xmath18 with side - length @xmath19 .
let @xmath20 and @xmath21 .
we denote by @xmath22 the area measure and by @xmath23 the cardinality of a finite set @xmath24 .
given @xmath25 , we say that a unit square @xmath16 ( i.e. , the square with @xmath26 ) is @xmath27-_good _ if ( i ) @xmath28 and ( ii ) @xmath29 . for any square @xmath15 , we put @xmath30 [ lemma - levsasha ] given @xmath31 there exists @xmath32 such that for any square @xmath15 with @xmath33 and any non - negative subharmonic function @xmath17 on a neighbourhood of @xmath15 with @xmath34 , @xmath35 it is instructive to juxtapose this estimate with a less restrictive one ( which also will be used below ) , where we require that _ almost every _ unit square @xmath36 contains a non - negligible piece of the set @xmath37 and get a much faster growth of @xmath17 .
[ lemma - adi ] let @xmath17 be a non - negative subharmonic function on a neighbourhood of the square @xmath15 with @xmath38 and let @xmath39 be a positive parameter .
suppose that for some @xmath25 and for all , except of at most @xmath40 unit squares @xmath36 , we have @xmath41 .
then , @xmath42 ^ 2\bigr)\ ] ] with some @xmath43 , provided that the size @xmath44 of the square @xmath15 is sufficiently large . note that our reduction of the first part of theorem [ thm1 ] to lemma [ lemma - levsasha ] is based on the pointwise ergodic theorem , and that the proof of lemma [ lemma - levsasha ] makes use of lemma [ lemma - adi ] . [ [ section-2 ] ] a natural idea to construct a non - trivial translation - invariant probability measure on @xmath0 ( and , in particular , for the proof of the second part of theorem [ thm1 ] ) is to use the classical krylov - bogolyubov construction .
we take a function @xmath45 , denote by @xmath46 the point mass on @xmath9 ( viewed as a probability measure on @xmath0 ) and average it along the orbit of @xmath8 defining @xmath47 in other words , for any borel set @xmath48 , @xmath49 then , we let @xmath50 , and consider the limiting measure .
the problem with this idea is that the space @xmath0 is not compact ; therefore , we need to ensure tightness of the family @xmath51 .
in addition , we must ensure that ( at least a part of ) the limiting measure is not supported by the constant functions .
thus , the entire function @xmath9 , which we start with , should be carefully chosen .
first , we construct a particular subharmonic function @xmath17 which can be thought as a certain approximation to @xmath52 .
we define a special unbounded closed set @xmath53 which can be thought as a two - dimensional fat cantor - type set viewed from the inside - out and a subharmonic function @xmath17 of a nearly minimal growth outside @xmath54 ( lemma [ lemma - sh ] ) .
then , using hrmander s classical estimates of solutions to @xmath55-equations , we build an entire function @xmath56 of a nearly minimal growth outside @xmath54 with needed properties ( lemma [ lemma - ef ] ) .
the functions @xmath17 and @xmath56 enjoy an interesting dynamical behaviour , and their construction is likely of independent interest .
[ [ section-3 ] ] we say that an entire function @xmath9 is _ locally uniformly recurrent _ if for every @xmath12 and every compact set @xmath57 the set @xmath58 is _ relatively dense _ in @xmath3 ( that is , any disk of sufficiently large radius contains at least one point of this set ) .
this is a locally uniform counterpart of bohr s classical definition of almost - periodicity . in @xcite ,
weiss gave a simple construction of functions of this class based on the runge approximation theorem .
locally uniformly recurrent entire functions can serve as a starting point for the krylov - bogolyubov - type construction described above . however , as the following theorem shows their growth is rather far from the minimal one .
[ thm - unform - recurrent ] \(a ) for any non - constant locally uniformly recurrent entire function @xmath9 , @xmath59 \(b ) there exists a non - constant locally uniformly recurrent functions @xmath9 such that @xmath60 note that the difference in the growth of entire functions in theorems [ thm1 ] and [ thm - unform - recurrent ] and that of the corresponding subharmonic functions in lemmas [ lemma - levsasha ] and [ lemma - adi ] are closely related . [ [ section-4 ] ] as we have already mentioned , translation - invariant probability measures on the space of entire functions must have heavy tails .
[ thm - heavytails ] \(a ) let @xmath7 be a non - trivial translation - invariant probability measure on the space of entire functions .
then , for every @xmath12 , @xmath61 = + \infty\,.\ ] ] \(b ) there exists a non - trivial translation - invariant probability measure @xmath7 on the space of entire functions such that , for every @xmath62 , @xmath63 here and elsewhere , the notation @xmath64 means that there exists a positive numerical constant @xmath65 such that @xmath66 .
[ [ section-5 ] ] it is natural to look at the counterparts of theorems [ thm1 ] and [ thm - unform - recurrent ] for meromorphic functions .
we treat meromorphic functions as maps of the complex plane into the riemann sphere endowed with the spherical metric @xmath67 , and denote by @xmath68 the space of meromorphic functions endowed with the topology of the locally uniform convergence in the spherical metric ( as usual , we treat @xmath69 as a constant meromorphic function ) . by @xmath4
we denote the borel sigma - algebra generated by this topology .
since @xmath70 , it is worthwhile to note that these definitions are consistent with the ones we have used above . to measure the growth of a meromorphic function @xmath9 we will use nevanlinna s characteristics @xmath71 .
it will be convenient to use it in the ahlfors - shimizu geometric form : @xmath72 where @xmath73 is the spherical derivative of @xmath9 .
then the inner integral in the definition of characteristics @xmath74 is the spherical area of the image of the disk @xmath75 considered with multiplicities of covering .
the basic properties of the nevanlinna s characteristics can be found , for instance , in ( * ? ? ?
* chapter 1 ) . here , we will mention that if @xmath9 is an entire function then the growth of its nevanlinna characteristics and of the logarithm of its maximum modulus are equivalent in the following sense : @xmath76 and , for every @xmath77 , @xmath78 we also point out that it is easy to see that if @xmath9 is a non - constant doubly periodic meromorphic function , then the spherical area of the image @xmath75 counted with multiplicities has quadratic grows with @xmath79 , and therefore , @xmath71 has quadratic growth as well .
as above , @xmath3 acts on @xmath68 by translations .
we call the probability measure @xmath7 on @xmath80 translation - invariant if it is invariant with respect to this action . as above
, we call a translation - invariant measure @xmath7 non - trivial if the set of all non - constant functions has measure zero . examples of non - trivial translation - invariant probability measures can be easily constructed by averaging the translations of a doubly periodic function . in these examples , for @xmath7-a.e .
function @xmath81 , @xmath82 as @xmath50 .
the following theorem shows that one can not do better : [ thm - meromorh - measures ] let @xmath7 be a non - trivial translation - invariant probability measure on meromorphic functions .
then , for @xmath7-a.e
. function @xmath81 , @xmath83 we call a meromorphic function @xmath9 _ locally uniformly recurrent _ if for every @xmath12 and every compact set @xmath84 , the set @xmath85 is relatively dense in @xmath3 . here , as above , @xmath67 is the spherical distance .
it is easy to see that doubly periodic meromorphic functions are locally uniformly recurrent .
i.e. , there are plenty of locally uniformly recurrent meromorphic functions @xmath9 with @xmath82 as @xmath50 . as in the previous case
, this estimate can not be improved : [ thm - meromorh - recurrent ] let @xmath9 be a non - constant locally uniformly recurrent meromorphic function
. then @xmath83 the authors are grateful to alexander borichev , gady kozma , fedor nazarov and benjy weiss for several very helpful discussions .
we thank steven britt for his assistance with language editing .
in this section the squares denoted by @xmath15 , @xmath86 , @xmath87 , and @xmath16 have vertices with integer - valued coordinates , @xmath15 is a square with large side - length @xmath38 , and @xmath17 is a subharmonic function on a neighbourhood of @xmath15 . by @xmath88
we denote the zero set of @xmath17 . assuming that for all but @xmath40 unit squares @xmath36 we have @xmath28 , we need to show that @xmath89 ^ 2 } u\,.\ ] ] first , we observe that if the disk @xmath90 centered at @xmath91 contains a portion of the zero set @xmath37 of area at least @xmath27 ( with @xmath92 ) , then @xmath93 whence , @xmath94 let @xmath95 be the integer part of @xmath96 .
put @xmath97 , take the points @xmath98 , @xmath99 , , @xmath100 , with @xmath101 , so that @xmath102 , @xmath103 , and consider the disks @xmath104 with sufficiently large integer @xmath105 .
we call the index @xmath106 _ normal _ if the disk @xmath107 contains at least one non - exceptional unit square @xmath16 . for normal indices
@xmath108 , we have @xmath109 if the disks @xmath110 ,
, @xmath111 are not normal , then the number of different exceptional squares contained in their union @xmath112 is @xmath113 . since the total number of exceptional unit squares does not exceed @xmath40 , we conclude that the number of not normal disks is @xmath114 provided that @xmath115 was chosen much larger than @xmath116 .
we conclude that there are at least @xmath117 indices @xmath118 , for which estimate holds .
hence , the lemma follows .
@xmath119 recall that we say that a unit square @xmath36 is @xmath27-good if @xmath28 and @xmath120 , and that for any square @xmath15 , we put @xmath121 our aim is to show that @xmath122 with no loss of generality we assume that @xmath123 for an integer @xmath124 ( then , @xmath125 ) .
we construct a sequence of squares @xmath126 , , @xmath127 , with @xmath128 .
first , we split the square @xmath129 into @xmath130 squares @xmath87 with @xmath131 . for these squares
@xmath87 we write @xmath132 , and note that @xmath133 then , according to certain rules described below , we choose one of the squares @xmath87 , and call it @xmath86 .
suppose that the squares @xmath134 ,
, @xmath129 have already been chosen .
we will fix the parameters @xmath135 and @xmath136 to be chosen later , and consider three cases . : _ there exist at least @xmath137 squares @xmath132 such that @xmath138_. we claim that in this case there exists at least one square @xmath132 with @xmath139 indeed , otherwise , gives us @xmath140 arriving at a contradiction .
then , we let @xmath86 be one of the squares @xmath132 such that holds . :
_ for all squares @xmath132 contained in the square @xmath141 _
, @xmath142 here , @xmath143 denotes the square with the same center as @xmath129 and @xmath144 .
we claim that if @xmath145 is chosen sufficiently small , then holds for at least one of the remaining squares .
otherwise , @xmath146 provided that @xmath147 .
as in the first case , we let @xmath86 be one of the squares @xmath132 such that holds .
we now consider the remaining case , which is complementary to the cases 1 and 2 : : _ there exists at least one square @xmath132 contained in @xmath143 such that @xmath148 ( with @xmath149 chosen above)_. at the same time , the number of squares @xmath150 with @xmath151 is not less than @xmath152 .
then we call one one these squares @xmath86 .
we also know that _ for at most @xmath153 squares @xmath150 , we have @xmath154_. now , we are ready to prove lemma [ lemma - levsasha ] .
first , we note that on each step the value @xmath155 either increases ( cases 1 and 2 ) , or decreases by a factor of at most @xmath156 . since the total number of steps is @xmath157
, we conclude that _ for each @xmath108 , @xmath158_. next , we observe that if on the @xmath108th step one of the cases 1 or 2 occurs , then by , @xmath159 will increase by a factor of at least @xmath160 .
since on other steps @xmath155 decreases not more than a factor of @xmath156 , choosing @xmath161 sufficiently large , ensures us that out of the @xmath124 steps at least @xmath162 steps result in case @xmath163 . ] the squares @xmath129 and @xmath86 in the 3rd case .
assume that on the @xmath108th step the 3rd case happens .
then , applying lemma [ lemma - adi ] ( with an appropriate scaling ) to the square @xmath164 with @xmath165 centered at the same point as @xmath86 ( and therefore , contained in @xmath129 ) , we obtain @xmath166 with some @xmath167 . since this happens for at least @xmath162 indices @xmath108 , we conclude that @xmath168 it remains to recall that @xmath125 and that , since @xmath15 was a good square , @xmath169 .
in this section , @xmath170 denotes the square of side - length @xmath67 centered at @xmath91 , and we let @xmath171 . we will be using a simple and known fact from the integral geometry : [ lemma - ig ] for any measurable set @xmath172 and any @xmath173 , @xmath174 [ [ section-6 ] ] applying the ergodic decomposition theorem ( see , for instance , ( * ? ? ?
* sections 6.1 and 8.6 ) ) , we can find a borel probabity space @xmath175 and a measurable map @xmath176 for which \(i ) for @xmath177-a.e .
@xmath178 , @xmath179 is a probability measure on @xmath5 , which is invariant and ergodic with respect to the action of @xmath3 on @xmath5 by translations @xmath8 ;
\(ii ) for every borel set @xmath180 , @xmath181 it is not difficult to see that the set of entire functions @xmath9 such that , for every @xmath12 , @xmath182 is a borel set .
hence , proving theorem 1a , it suffices to assume that the measure @xmath7 is _ ergodic _ with respect to translations @xmath8 .
[ [ section-7 ] ] put @xmath183 .
given @xmath184 , consider the borel sets @xmath185 for @xmath186 , we have @xmath187 , @xmath188 .
we denote by @xmath189 , @xmath188 , the corresponding limiting sets as @xmath190 . since the complement @xmath191 consists of constant functions and the measure @xmath7 does not charge constants , @xmath192 .
[ [ section-8 ] ] we claim that @xmath193 as well .
otherwise , by translation - invariance of the set @xmath194 and by ergodicity of @xmath7 , we have @xmath195 . applying lemma [ lemma - ig ]
, we conclude that for @xmath7-a.e .
@xmath9 and for every @xmath196 , @xmath197 then , by the wiener version of the pointwise ergodic theorem ( see for instance , @xcite ) for every @xmath184 , @xmath198 i.e. , for @xmath7-a.e .
@xmath45 , @xmath199 , which means that @xmath200 everywhere in @xmath3 .
that is , @xmath9 a constant function , which is a contradiction .
[ [ section-9 ] ] now , we fix @xmath184 so that @xmath201 , and let @xmath202 we claim that _ for @xmath7-a.e .
@xmath45 , the limit @xmath203 exists and is @xmath204_. indeed , for any @xmath45 and any @xmath205 , by lemma [ lemma - ig ] , we have @xmath206 and by the pointwise ergodic theorem , for @xmath7-a.e .
@xmath9 , the @xmath50 limit of the rhs exists and equals @xmath207 applying fubini s theorem and then using the translation - invariance of the measure @xmath7 , we can rewrite this expression as @xmath208\ , { \rm d}a(z ) \\ \\
& \qquad = \frac1{a(s_r)}\,\int_{s_r } \bigl[\ , \int_{\mathcal e}\done_{x(\tau_z f , \rho)}(0)\ , { \rm d}{\lambda}(f ) \,\bigr]\ , { \rm d}a(z ) \\ \\ & \qquad = { \lambda}\bigl\{f\in{\mathcal e}\colon 0\in x(f , \rho ) \bigr\ } \\ \\ & \qquad = { \lambda}\bigl ( { \mathcal e}_1(\rho ) \cap { \mathcal e}_2(\rho ) \bigr ) { \geqslant}\frac9{10}\,,\end{aligned}\ ] ] proving the claim . [
[ section-10 ] ] it remains to show that if @xmath9 is a non - constant entire function such that for some @xmath184 , @xmath209 then , for every @xmath12 , @xmath210 first , we note that it suffices to show that holds for the sequence @xmath211 ; then the general case follows .
then , we take @xmath212 with sufficiently large @xmath213 , split the square @xmath214 into @xmath215 squares @xmath16 squares @xmath216 with side - length @xmath217 , and consider the subharmonic function @xmath218 . by the last claim , for at least half of the squares @xmath216 , @xmath219 and @xmath220 .
applying lemma [ lemma - levsasha ] , we complete the proof .
@xmath119 note that with a little effort one can extract from lemma [ lemma - levsasha ] slightly more than theorem 1a asserts , namely , that for @xmath7-a.e .
@xmath45 , @xmath221 likely , this estimate can be somewhat improved .
the proof is straightforward .
let @xmath9 be a non - constant locally uniformly recurrent function , and let @xmath222 ^ 2}|f|$ ] . applying the definition of locally uniform recurrency with @xmath223 ^ 2 $ ] and @xmath224
, we see that there exists @xmath225 such that for every square @xmath15 with the side - length @xmath44 , @xmath226
. then , lemma [ lemma - adi ] does the job . @xmath119
we will use a version of the classical carleman - levinson - sjberg loglog - theorem , cf .
likely , this lemma can be deduced from at least one of many known versions of the loglog - theorem . since
its proof is quite simple , for the reader s convenience , we will supply it .
[ lemma - loglog ] suppose @xmath17 is a non - constant subharmonic function in @xmath3 . then
, for every @xmath12 , @xmath227 this lemma immediately yields theorem 3a : applying , as above , the ergodic decomposition theorem , we may assume that the measure @xmath7 is ergodic . then , by the pointwise ergodic theorem , for @xmath7-a.e .
entire function @xmath9 we have @xmath228\,,\ ] ] and since the measure @xmath7 does not charge constant functions , by lemma [ lemma - loglog ] , the limit on the rhs is infinite .
@xmath119 we let @xmath229 and choose @xmath95 so that @xmath230 , with some @xmath231 to be chosen . for @xmath232 , we take @xmath233 , @xmath234 , so that @xmath235 and let @xmath236
. then , we put @xmath237 , @xmath232 , let @xmath107 be the disks centered at @xmath233 of radius @xmath238 , and let @xmath239 .
note that the sets @xmath240 are disjoint and that @xmath241 .
we claim that * _ if @xmath242 is chosen sufficiently close to @xmath243 and @xmath244 is sufficiently small , then for every @xmath232 , @xmath245 on a subset @xmath246 with @xmath247_. indeed , let @xmath248 . then , @xmath249 whence @xmath250 provided that @xmath251 and @xmath252 . it remains to put @xmath253 .
now , @xmath254 let @xmath255 .
then , for @xmath256 , @xmath257 letting @xmath258 , we get @xmath259 then , letting @xmath260 , we conclude the proof .
both proofs are quite straightforward . as in the previous proofs
we may assume that the measure @xmath7 is ergodic .
then , by the pointwise ergodic theorem , for @xmath7-a.e .
meromorphic function @xmath9 , @xmath261\,.\ ] ] since @xmath7-a.s .
, the function @xmath9 is not a constant ( and the distribution of @xmath7 is translation invariant ) , the rhs is positive ( may be infinite ) .
thus , for sufficiently large @xmath262s , @xmath263 and therefore , @xmath264 .
@xmath119 let @xmath9 be a non - constant locally uniformly recurrent meromorphic function .
we fix a disk @xmath265 such that @xmath9 _ is analytic _ on @xmath266 , take the closed spherical disk @xmath267 such that @xmath268 , and denote by @xmath269 the spherical distance between @xmath270 and the curve @xmath271 . by the definition of local uniform recurrency , each square @xmath15 with sufficiently large length - side @xmath19 contains a point @xmath272 such that @xmath273 , where @xmath67 is the spherical metric .
denote by @xmath274 the disk centered at @xmath272 of the same radius as @xmath265 .
we claim that @xmath275 . to show this , fix a point @xmath276 .
then , by the argument principle , the index of the curve @xmath271 with respect to the point @xmath277 is positive .
furthermore , when the point @xmath91 traverses the circumference @xmath278 , @xmath279 traverses the curve @xmath271 , @xmath280 traverses the curve @xmath281 , and the spherical distance between @xmath279 and @xmath280 remains less than @xmath282 , while @xmath283 .
hence , the index of the curve @xmath281 with respect to @xmath277 coincides with that of @xmath271 , and therefore , is positive as well .
thus , @xmath284 proving the claim .
denote by @xmath285 the square having the same center as @xmath15 and with the length - side @xmath286 . since @xmath287
, we conclude that @xmath288 .
hence , the spherical area of @xmath289 is not less than that of @xmath290 .
packing the disk @xmath291 with sufficiently large @xmath262 by about @xmath292 disjoint translations of the square @xmath285 , we see that the spherical area of @xmath293 is @xmath294 , which yields the theorem .
we will construct the closed set @xmath53 which we will call _ the ternary systems of squares_. it will be defined as the limit of the increasing sequence @xmath295 of compact sets such that @xmath296 consists of @xmath297 and its eight disjoint translations .
one can think about the limiting set @xmath54 as a two - dimensional ternary cantor - type set viewed from the inside - out . for @xmath172 and @xmath298
, we put @xmath299 here and elsewhere , @xmath300 denotes the @xmath301-distance on @xmath302 . for @xmath172 , we put @xmath303 . that is ,
if the function @xmath304 is defined on @xmath24 , then @xmath305 is defined on @xmath306 .
we fix a sequence @xmath307 and define : * the increasing sequence @xmath308 by @xmath309 , @xmath310 ; * the squares @xmath311 ^ 2 $ ] ; * the translates @xmath312 , where @xmath313 ; see figure [ fig : squares ] . ]
the square @xmath314 with 9 copies of the square @xmath315 . then , @xmath316 , @xmath317 , and finally , @xmath318 . for every @xmath319 ,
the set @xmath296 consists of @xmath320 disjoint copies of @xmath321 .
we denote by @xmath322 , @xmath323 , the centers of these copies .
that is , there exist indices @xmath324 such that @xmath325 ( in particular , @xmath326 ) , and @xmath327 next , we denote by @xmath328 the union of the corridors left on the @xmath213th step of the construction and the outer perimeter corridor that goes along the boundary @xmath329 ( see figure [ fig : corridors ] ) . ] the corridors @xmath328 .
the width of these corridors is @xmath330 , where @xmath331 .
that is , @xmath332 to simplify computations , in what follows , we always assume that @xmath333 and that @xmath334 . since @xmath335 these assumptions yield that @xmath336 in particular , the sequence @xmath337 is increasing .
we will call the set @xmath54 _ fat _ if @xmath338 . in this case , @xmath339 with @xmath340 .
note that since @xmath296 consists of @xmath341 disjoint translations of the square @xmath342 , @xmath343 .
if the set @xmath54 is fat then @xmath344 , and @xmath345 with some @xmath346 .
in particular , fat sets @xmath54 ( and only they ) have positive relative area . given a ternary system of squares @xmath54
, we call the function @xmath347 _ concordant _ with @xmath54 if for every @xmath348 and @xmath323 , @xmath349 given a sequence @xmath350 , we say that the function @xmath351 is @xmath269-concordant with @xmath54 if for every @xmath348 and @xmath323 , @xmath352 for a continuous function @xmath351 and a compact set @xmath2 , we put @xmath353 define _ the majorant _ @xmath354 with sufficiently large positive @xmath4 and put @xmath355 . then , define the sequence @xmath356 by @xmath357 [ lemma - ef ] for any sufficiently large positive @xmath4 , there exists a non - constant entire function @xmath56 which is @xmath356-concordant with @xmath54 , satisfies @xmath358 and @xmath359 we start with the subharmonic counterpart of this lemma .
[ lemma - sh ] for any @xmath360 , there exists a sequence of non - negative subharmonic functions @xmath361 in @xmath3 with the following properties : \(i ) for each @xmath362 , @xmath363 \(ii ) @xmath364 ; \(iii ) @xmath365 .
note that by property ( i ) , for every @xmath366 , @xmath367 everywhere on the square @xmath314 .
hence , the sequence @xmath368 converges to the limiting subharmonic function @xmath17 . by property ( i ) , the limiting function is concordant with @xmath54 . by ( ii ) , we have @xmath369 , @xmath370 , and by ( iii ) , @xmath371 , @xmath372 .
_ proof of lemma [ lemma - sh ] _ : take the subharmonic function @xmath373 scale it @xmath374
@xmath375 ] scaling , shifting and rotating the function @xmath376 . and
take the upper envelope of @xmath377 shifted and rotated copies of @xmath378 : @xmath379 we will need two estimates : @xmath380 and @xmath381 we put @xmath382 .
assuming that the subharmonic functions @xmath383 have been already defined , we glue together the functions @xmath384 , putting @xmath385 where @xmath386 is the subharmonic function defined in . note that this definition ensures property ( i ) in the statement of the lemma , as @xmath387 we claim that , for @xmath388 and @xmath372 , @xmath389 ( with @xmath355 ) . this claim yields that @xmath390 and therefore , the functions @xmath361 , @xmath372 , are subharmonic in @xmath3 .
the case @xmath391 of our claim follows from the lower bound @xmath392 on @xmath393 .
now , let @xmath394 .
we know that @xmath395 outside the set @xmath396 .
note that @xmath397 hence , @xmath395 on @xmath398 .
furthermore , by the bound , on @xmath398 we have @xmath399 . for @xmath400
, we have @xmath401 , whence @xmath402 on @xmath398 . on the other hand , @xmath403 , so applying the lower bound for @xmath386 , we get the claim
. note that @xmath404 on @xmath405 , and by , @xmath406 therein . this proves ( ii ) .
at last , on @xmath407 , we have @xmath408 proving ( iii ) .
@xmath119 we put @xmath409 and construct a sequence @xmath410 of entire functions with the following properties : \(i ) for @xmath394 and @xmath411 , @xmath412 \(ii ) for @xmath394 , @xmath413 then , the existence of the function @xmath56 will follow from the following claim .
[ claim2 ] for every @xmath414 , @xmath415 first , assuming that estimates ( i ) and ( ii ) and the claim hold , we complete the proof of lemma [ lemma - ef ] . on the second step ,
we prove the claim assuming that the property ( i ) holds . on the last step
, we construct the sequence @xmath410 having properties ( i ) and ( ii ) .
we put @xmath416 by ( i ) , the series converges locally uniformly in @xmath3 .
moreover , @xmath417 and then , for @xmath418 , @xmath419 combining these inequalities with the claim , we conclude that , for every @xmath420 , @xmath421 provided that the parameter @xmath4 is large enough .
that is , the limiting entire function @xmath56 is @xmath356-concordant with @xmath54 .
furthermore , by properties ( i ) and ( ii ) , the function @xmath56 satisfies @xmath422 and @xmath423 provided that @xmath4 is sufficiently large .
we use induction on @xmath424 .
the base of induction @xmath425 is exactly property ( i ) .
now , assuming that the claim holds for the pair @xmath426 , we will prove it for @xmath427 .
for every @xmath428 , we write @xmath429 and put @xmath430 then we have @xmath431 and @xmath432 then , taking into account that @xmath433 and using ( b ) , we get @xmath434 @xmath435 now , adding ( a ) and ( c ) , we conclude proof of the claim
. @xmath119 thus , it remains to construct a sequence of entire functions @xmath410 satisfying conditions ( i ) and ( ii ) .
we fix a sequence of smooth cut - off functions @xmath436 , @xmath437 , so that @xmath438 and @xmath439 ( such a sequence exists since @xmath440 ) .
we put @xmath409 and suppose that the functions @xmath441 ,
, @xmath442 have already been constructed .
we put @xmath443 and note that @xmath444 with @xmath445 .
then , we define the function @xmath446 by @xmath447 , where @xmath448 is hrmander s solution ( * ? ? ?
* theorem 4.2.1 ) to the @xmath55-equation @xmath449 satisfying @xmath450 where @xmath361 are the subharmonic functions constructed in lemma [ lemma - sh ] .
note that @xmath451 and therefore , @xmath452 on @xmath453 .
furthermore , since @xmath454 , we have @xmath455 with @xmath456 ( in the second inequality we have used the inductive assumption ) .
taking into account that the area of @xmath457 is less than @xmath458 and recalling that @xmath459 on @xmath460 , we conclude that the integral on the rhs of does not exceed @xmath461 provided that the constant @xmath4 is sufficiently large .
therefore , by hrmander s theorem , @xmath462 here , we aim to show that @xmath463 let @xmath464 be a positive constant . then , for @xmath465 , we have @xmath466 to estimate the first integral , we observe that @xmath467 ( in the second inequality we have used the induction assumption ) .
thus , @xmath468 provided that the constant @xmath4 is sufficiently large . to estimate the second integral , using the fact that @xmath465 , we write @xmath469
whence , @xmath470 again , provided that the constant @xmath4 is large enough .
thus , @xmath471 everywhere on @xmath472 , as we have claimed .
first , we note that @xmath473 and that @xmath448 is analytic in the @xmath474-neighbourhood of each of the sets @xmath475 .
then , for every @xmath362 and every @xmath476 , we have @xmath477 recall that @xmath478 on each square @xmath475 .
then , recalling that @xmath479 and choosing @xmath480 , we see that @xmath481 therefore , @xmath482 provided that @xmath4 is sufficiently large , and finally , @xmath483 again , provided that @xmath4 is sufficiently large .
this completes the ( somewhat long ) proof of lemma [ lemma - ef ] .
in this section , we denote by @xmath484 any increasing sequence of squares centered at the origin with the side - lengths tending to infinity .
if @xmath485 is a square and @xmath172 is a borel set , then we denote the relative area of @xmath24 in @xmath16 by @xmath486 for an entire function @xmath487 , let @xmath488 denote its orbit and @xmath489 denote the closure of @xmath490 in @xmath0 . for a compact set @xmath57 and a continuous function @xmath491 , we denote by @xmath492 , the oscillation of @xmath304 on @xmath2 .
[ lemma - kb ] let @xmath493 .
\(i ) suppose that there exists an increasing sequence @xmath494 such that @xmath495 and there exists a square @xmath16 and a constant @xmath496 such that @xmath497 then there exists a translation - invariant probability measure @xmath7 supported by @xmath489 which does not charge the constant functions .
\(ii ) furthermore , suppose that condition is replaced by a stronger one : @xmath498 and that condition continues to hold .
then , for @xmath7-a.e .
@xmath45 , @xmath499 it is worth mentioning that condition ( i ) already yields the upper bound though only on a subsequence of the squares @xmath500 : for @xmath7-a.e .
@xmath45 , @xmath501 consider the sequence of probability measures on @xmath0 : @xmath502 in other words , for any borel set @xmath503 , @xmath504 we claim that @xmath505 _ is a tight sequence _ of probability measures , that is , for every @xmath506 , there exists a compact set @xmath507 such that , for every @xmath372 , @xmath508 . to see this , given @xmath157
, we choose @xmath509 so that for @xmath510 , @xmath511 and let @xmath512 where @xmath513 is the square concentric with @xmath514 and having double the side - length .
the sets @xmath515 are compact subsets of @xmath0
. we will show that for any @xmath372 and any @xmath516 , @xmath517 which yields the tightness of @xmath505 .
indeed , @xmath518 where @xmath519 , and @xmath520 . for @xmath521 and @xmath522 , we have @xmath523 ( recall that @xmath524 ) , whence , for these @xmath213s , @xmath525 . on the other hand , for @xmath510 , we have @xmath526 thus , @xmath527 proving the tightness of @xmath505
. now , let @xmath7 be any limiting probability measure for the sequence @xmath505 .
since each measure @xmath528 is supported by the orbit @xmath490 , clearly , @xmath7 is supported by the closure of the orbit @xmath489 .
the measure @xmath7 is translation - invariant .
this follows from the fact that for any @xmath372 , any @xmath529 , and any borel set @xmath503 , @xmath530 where @xmath531 denotes the symmetric difference of sets , and @xmath532 is the side - length of @xmath314 . at last
, we can specify the measure @xmath7 such that it will not charge the set @xmath533 of constant functions .
indeed , following our assumption and passing if necessary to some subsequence , we may assume that a positive limit exists @xmath534 this yields that @xmath535 . to see this ,
let @xmath536 .
then @xmath537 is an open set and @xmath538 .
hence , it is enough to show that , for each @xmath213 , @xmath539 .
this holds since @xmath540 then , if needed , we replace @xmath7 by its restriction on @xmath541 and normalize it to make @xmath7 the probability measure .
this completes the proof of part ( i ) .
@xmath119 now , we suppose that condition holds , and assume that the probability measures @xmath505 and @xmath7 are the same as in the proof of part ( i ) .
consider the open set @xmath542 we have @xmath543 whence , by , @xmath544 furthermore , since the sets @xmath545 are open , @xmath546 so @xmath547 hence , applying the borel - cantelli lemma , we conclude that @xmath548 which means that @xmath7-a.e
. @xmath45 does not belong to any @xmath545 with @xmath549 , i.e. , @xmath550 this proves part ( ii ) and finishes off the proof of lemma [ lemma - kb ] .
after the work we have done in lemmas [ lemma - ef ] and [ lemma - kb ] , the proofs of these theorems is rather straightforward .
we take the sequence @xmath551 put @xmath309 and @xmath552 for @xmath372 , and ( with some conflict of notation used in lemma [ lemma - sh ] ) @xmath553 ^ 2 $ ] . by @xmath56
we denote the corresponding entire function with properties as in lemma [ lemma - ef ] .
we fix a sufficiently large value of the parameter @xmath4 as in lemma [ lemma - ef ] and then will drop dependence on @xmath4 from our notation . we claim that first , we verify convergence of the series @xmath556 for this , we need to bound the relative area @xmath557 we note that for @xmath558 ^ 2 $ ] , the translations @xmath559 belong to @xmath560 .
thus , for @xmath561 , @xmath562 , we have @xmath563 the relative area of the set of these @xmath272s in @xmath564 is @xmath565 hence , @xmath566 at last , for @xmath567 , @xmath568 with some @xmath340 , and @xmath569 therefore , the rhs of is @xmath570 which is what we need for condition . to verify condition ,
we take @xmath571 ^ 2 $ ] , @xmath572 , and note that for @xmath573 with @xmath574 , @xmath575 , we have @xmath576 ^ 2 $ ] . therefore , @xmath577 ^ 2 } |g| \\ & { \geqslant}{\operatorname{osc}}_{[-1 , 1]^2}|g| -\delta_1 \\ & { \geqslant}{\operatorname{osc}}_{[-1 , 1]^2}|z| - \frac13 - \delta_1 \\ & { \geqslant}c > 0\end{aligned}\ ] ] since @xmath578 ^ 2 } |z|$ ] .
thus , the set @xmath579 contains the @xmath269-neighbourhood of the set @xmath580 .
hence , the relative area of this set in @xmath554 is bounded from below by @xmath581 which yields condition . at last ,
applying lemma [ lemma - kb ] , we see that for @xmath7-a.e .
@xmath45 , @xmath582 ^ 2 } \bigl ( |f| - \exp \mathcal m(2^{n+1 } ) \bigr ) { \leqslant}1\,.\ ] ] in our case @xmath583 , whence @xmath584 . then , given @xmath585 , we choose @xmath213 such that @xmath586 and get @xmath587 ^ 2 } \log |f| { \leqslant}\exp(c 2^{2n } n^3 ) \,.\ ] ] furthermore , recalling that @xmath588 , we see that @xmath589 , whence @xmath590 , proving theorem 1b .
@xmath119 here , we take @xmath591 , and let @xmath56 be the entire function constructed by using lemma [ lemma - ef ] . note that in this case @xmath592 given @xmath585 we choose @xmath213 so that @xmath593 and get @xmath594 furthermore , given a square @xmath500 , for any @xmath595 and any @xmath596 , we have @xmath597 given @xmath12 and a compact set @xmath57 , we choose @xmath124 so large that @xmath598 and @xmath599 observing that each square @xmath600 with the side length @xmath601 contains at least one point of the set @xmath602 we complete the proof of theorem 2b .
@xmath119 as in the proof of theorem 2b , we take @xmath591 .
we denote by @xmath484 the corresponding ternary system of squares and let @xmath56 be the entire function as in lemma [ lemma - ef ] .
we fix @xmath4 so large that @xmath603 and drop the parameter @xmath4 in our notation . as in the proof of theorem 1b ,
a straightforward verification , which we skip , shows that conditions and of lemma [ lemma - kb ] are satisfied .
we fix @xmath607 sufficiently large and choose @xmath124 so that @xmath608
. then for all @xmath607 s ( except maybe a countable set of values which we may neglect ) , @xmath609 since @xmath610 , we have @xmath611 on @xmath612 , as well as on all translations @xmath613 .
thus , @xmath614 on the other hand , we have @xmath615 whence @xmath616 completing the proof of theorem 3b . @xmath119 | we study non - trivial translation - invariant probability measures on the space of entire functions of one complex variable .
the existence ( and even an abundance ) of such measures was proven by benjamin weiss . answering weiss question , we find a relatively sharp lower bound for the growth of entire functions in the support of such measures .
the proof of this result consists of two independent parts : the proof of the lower bound and the construction , which yields its sharpness .
each of these parts combines various tools ( both classical and new ) from the theory of entire and subharmonic functions and from the ergodic theory .
we also prove several companion results , which concern the decay of the tails of non - trivial translation - invariant probability measures on the space of entire functions and the growth of locally uniformly recurrent entire and meromorphic functions . |
regression models are one of the main tools of statistical modeling and supervised machine learning . in regression models ,
_ responses _ are related to _ predictors _ by a probabilistic model that is a function of model parameters that we want to estimate from the data . the most common regression model , linear regression , assumes the response follows a gaussian distribution that can take values anywhere in the real numbers
however , different applications have a response of a different nature . for example
, the response might take values only in the positive real numbers ( e.g. , the time between the arrivals of two buses ) , or in the non - negative integers ( e.g. , the number of thunders in a day , or the number and identity of the items chosen by someone from a catalog of items ) . in such cases , assuming that the response is gaussian might produce an inferior model to one obtained assuming a different distribution that better matches the nature of the response data . when the response takes on two values , it may be modeled as a bernoulli random variable . when it is a non - negative real number , it may be modeled as an exponential , erlang , or gamma distribution .
when the response is a non - negative integer it may be modeled as a poisson , negative binomial , or binomial random variable .
when the response is a category it may be modeled by a multinomial or categorical distribution .
all these distributions , and more , are part of the so - called exponential family ( @xcite , @xcite , @xcite ) .
the exponential family also includes distributions for random vectors with vector entries that are correlated , e.g. , as in the multivariate gaussian distribution , as well as with independent entries with different distribution types .
the generalized linear model ( glm ) introduced in @xcite provides the theory to build regression models with static parameters where the response follows a distribution in the exponential family .
the glm can then be seen as a generalization of linear regression .
many applications of regression models , indeed those that rely on the glm , assume the parameters are static .
this assumption is too restrictive when modeling time - series , which often exhibit trends , seasonal components , and local correlations .
similarly , in many applications the model parameters represent variables that describe a portion of the state of the underlying system that can not be or are not directly measured , but with well - known relationships that describe their dynamics .
the response can then be thought of as a noisy function of the system state and the predictors , and the goal of the regression model is to estimate the system state over time based on the observation time - series .
situations where an underlying dynamical system is only observed through noisy measurements are often encountered in engineering applications , and the celebrated and widely used kalman filter , introduced in @xcite solves the corresponding regression model when the parameter dynamics ( linear with additive gaussian noise ) and the observation distribution ( gaussian with a mean that is a linear function of the state ) are simple enough .
the kalman filter can be seen as an algorithm that efficiently estimates the mean and covariance matrix of the parameters in a linear regression model , where the parameters evolve in time through linear dynamics , based on the time - series of responses and predictors .
furthermore , the kalman filter is an online algorithm that updates the parameter estimates in a relatively simple fashion when new observations arrive , without having to re - analyze all previous data .
in addition , it is a second - order algorithm , in contrast to stochastic gradient descent ( e.g. , @xcite ) .
so while the kalman filter takes more computation per observation than stochastic gradient descent , it can converge to an accurate estimate of the parameters with fewer observations , while allowing for much more flexibility in the modeling of the underlying parameter dynamics .
lastly but importantly , because the kalman filter provides an estimate of the mean and covariance matrix , unlike other approaches that only focus on the mean , its results can be used to construct confidence intervals or samples of the parameters or of the model predictions these statistics are often necessary in several applications such as in contextual multi - armed bandits ( e.g. , @xcite ) .
many generalizations to the kalman filter have now been developed , e.g. , to allow for non - linear state dynamics state dynamics , or for observations that are noisy and non - linear and/or non - gaussian functions of the state as in the extended kalman filter , e.g. , see @xcite for a good overview .
more recently , there has been interest in merging the ideas from kalman filtering , namely modeling dynamics in the parameters of a regression model , with the flexibility to model observations from the wide class of probability distributions that the glm allows through its treatment of the exponential family ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
it turns out that approximate algorithms that are very similar to the extended kalman filter can be derived for a wide range of choices for the observation distributions . introducing and describing some of these algorithms for a fairly general class of models , including the multivariate glm , is the main focus of this paper .
the derivation we follow is novel and simpler ( e.g. , it does not invoke any kalman filter theory nor uses conjugacy ) . and
the form of the exponential family we use is slightly more general than that used in other references on these methods , because the nuissance parameter matrix @xmath0 included in our model is absent in other references that deal with multivariate responses .
the second focus of this paper is to propose the application of these methods to the contextual multi - armed bandit problem ( @xcite , @xcite ) , and show the resulting performance through simulations .
this application is novel to the best of our knowledge .
it broadens the class of models that can be used to describe the rewards , separates the concept of a response and a reward , and allows for the explicit modeling of dynamics in the parameters that map the context to the rewards .
section [ sec : setup ] introduces the class of models we study , and reviews the exponential family and the generalized linear model .
section [ sec : algo ] describes the general online algorithm to estimate the mean and variance of the parameters , and specializes it to the multivariate exponential family , the univariate exponential family , and to several examples of commonly used distributions .
we sketch the derivation of the algorithm in section [ sec : derivation ] .
we apply the methods developed to the contextual multi - armed bandit problem in sections [ contextualbandits ] , and conclude in section [ conclusion ] .
we assume all vector and matrix entries are real numbers .
we denote vectors using boldface small caps , and assume them to be column vectors .
we use small caps for scalars , and boldface large caps for matrices . if @xmath1 is a matrix , we denote its inverse by @xmath2 , and its transpose by @xmath3 .
we assume we have received @xmath4 pairs of observations @xmath5 , for @xmath6 where @xmath7 is the @xmath8-th _ response _ that we want to explain based on the @xmath8-th _ predictor _ @xmath9 , for some positive integers @xmath10 , @xmath11 and @xmath12 .
we denote by @xmath13 the information or history after @xmath4 observations , i.e. , @xmath13 is the set that includes the first @xmath4 responses and predictors . we postulate a model that relates the response to the predictor via a probability distribution @xmath14 where @xmath15 are the model parameters at the @xmath8-th observation , with one parameter for each row in the predictor matrix @xmath16 @xmath17 is a @xmath10-by-@xmath10 matrix and nuisance parameter that is assumed known and that we will often omit . as we will see later , the nuisance parameter plays the role of the covariance of the observations in linear regression , and is the identity matrix in many other cases of interest .
we consider regression models where the probability of the response depends on the predictors and the parameters only through the @xmath12-by-@xmath18 _ signal _ @xmath19 namely , models where @xmath20 rather than working with @xmath21 , we will typically work with its logarithm , denoted by @xmath22 which we assume to be a well - behaved function , particularly that the first and second derivatives with respect to @xmath23 exist and are finite .
many of the commonly used regression models fit the model above , including univariate and multivariate linear regression with response @xmath24 logistic or binomial , categorical or multinomial , exponential , poisson , negative binomial , gamma , etc .
our model also includes cases where the different entries in the response vector are conditionally independent given the signal , and follow any distribution that is a function of the signal or a subset of the signal entries .
for example , we can have as many predictor vectors as entries in @xmath25 , i.e. , @xmath26 and have the @xmath27-the response entry depend only on the @xmath27-th signal entry so that @xmath28 , with @xmath29 being a different function for different response entries @xmath27 . because all the entries still depend on the parameters @xmath30 this setup allows for combining the information from different types of measurements that depend on the parameters to obtain more accurate parameter estimates .
also , the predictor matrix @xmath31 may have a lot of structure , e.g. , to allow for some parameters to be shared across entries in the response vector , and others that are specific to subsets of the response .
here we drop the time subscript of our vectors and matrices to avoid notational clutter , so , e.g. , @xmath32 becomes simply @xmath25 .
all of the probability distributions of interest to model the response @xmath25 mentioned above , and more , can be re - arranged so that their log - likelihood has the following so - called _ natural exponential form _
@xmath33 here , @xmath34 is a @xmath10-by-1 vector referred to as the natural parameter , with @xmath35 in its @xmath27-th entry .
it is a function of the signal @xmath36 in our models this will be made specific in the next section .
crucially , the function @xmath37 is independent of @xmath25 and the function @xmath38 is independent of @xmath39 we assume that @xmath37 is twice differentiable with respect to its argument .
the @xmath10-by-@xmath10 nuisance parameter matrix @xmath0 is assumed symmetric and known .
when @xmath0 is unknown , it can be estimated through several methods that are not online , e.g. , see chapter 13 in @xcite .
it can be shown , e.g. , see appendix [ sec : meanvarnef ] , that the mean and covariance matrix of @xmath25 are given by @xmath40 = \mathbf{\phi } \frac{\partial b}{\partial \mathbf{\eta } } \label{eq : nefmean } \\ \mathbf{\sigma_y(\eta)}=&e[(\mathbf{y}- \mathbf{\mu(\eta)})(\mathbf{y}- \mathbf{\mu(\eta ) } ) ' ] = \mathbf{\phi } \frac{\partial^2 b}{\partial \mathbf{\eta}^2 } \mathbf{\phi } , \label{eq : nefvar}\end{aligned}\ ] ] where @xmath41 is a column vector with @xmath42 in its @xmath27-th entry , and @xmath43 is the @xmath10-by-@xmath10 matrix with @xmath44 in its @xmath8-th row and @xmath27-th column .
most of the frequently encountered univariate and multivariate distributions have the exponential form above .
in addition , a union of independent random vectors that are in the natural exponential family is also in the natural exponential family .
e.g. , a random vector with entries distributed according to different members of the exponential family is still in the natural exponential family with a natural parameter given by the union of the natural parameters of its entries .
this will allow us to estimate shared parameters in a regression model from multiple time series of a potentially different nature .
we consider models where the response is distributed according to equation [ eq : nef ] , which is a function of @xmath34 .
we use the glm to relate @xmath34 to the signal @xmath45 in the glm , introduced in @xcite , we assume that the signal @xmath36 is a function of the mean @xmath46 of the observation @xmath25 , in addition to assuming equation [ eq : nef ] for the observation @xmath25 .
the glm then assumes that there is a known one - to - one mapping between the natural parameter @xmath34 in equation [ eq : nef ] and the signal @xmath47 so we can view the likelihood or any statistic of @xmath25 either as a function of @xmath34 or of @xmath36 .
specifically , we have @xmath48 for known functions @xmath49 @xmath50 and @xmath51 , and where the so - called _ link function _
@xmath51 maps the mean @xmath46 of @xmath25 to the signal .
the mean @xmath46 and covariance @xmath52 in equations [ eq : nefmean ] and [ eq : nefvar ] can then be seen to be either a function of the natural parameter @xmath34 or of the signal @xmath53 e.g. , @xmath54 here the function @xmath55 is the inverse of @xmath51 .
we refer to @xmath56 as the _ response function _ ; it maps the signal to the mean of the response and plays a prominent role in our algorithms .
any invertible function @xmath51 can be used as the link function in a glm , but one that often makes sense , and where the mathematics to learn the model simplifies , is the _
canonical link _ that results in the signal being equal to the natural parameter , i.e. , in @xmath57 previous treatments of the glm in the context of dynamic parameters consider either a univariate response with the univariate case of equation [ eq : nef ] ( e.g. , see @xcite ) , or a multivariate response with equation [ eq : nef ] but with the nuissance parameter matrix @xmath0 equal to the identity ( e.g. , see chapter 2 in @xcite ) . in this sense
our treatment is a slight generalization .
we assume that the parameters evolve according to @xmath58 where @xmath59 is a zero - mean random vector with known covariance matrix @xmath60 we also assume that the _ noise _ @xmath61 is uncorrelated with itself over time , and uncorrelated with the observation parameters .
the known vector @xmath62 drives the parameters through the appropriately sized and known matrix @xmath63 and @xmath64 is a known @xmath11-by-@xmath11 square matrix .
we also assume that at time zero the mean and variance of @xmath65 are known , i.e. , that @xmath66 the general setup above includes several special cases of interest , described next .
when @xmath67 ( the identity matrix ) , @xmath68 , and @xmath69 we end up with the simple parameter dynamics @xmath70 which is the standard regression problem with static parameters .
this becomes the glm when we also assume that the response is in the exponential family , with natural parameter that is a function of the signal . in this sense ,
our model is a generalization of the glm where the parameters are allowed to vary over time .
when @xmath67 and @xmath68 , we end up with the simple parameter dynamics @xmath71 .
this allows the parameters to drift or diffuse over time in an unbiased ( zero - mean ) way , according to the noise @xmath61 .
this model is appealing for a range of applications , e.g. , it could model the conversion rate of visitors to the netflix signup page as a function of their browser , country , day of week , time of day , etc .
equation [ eq : pardynamics ] is central to the study of linear dynamical systems , control theory , and other related areas . specifically , it is core to the kalman filter , which also assumes a dynamical system that evolves according to equation [ eq : pardynamics ] , but with a response that is a linear function of the parameters ( the state in kalman filter parlance ) and additive gaussian noise .
the kalman filter is an online algorithm that estimates the mean and variance of the system state from the noisy observations .
so our setup is very related .
the main difference is that we do not restrict our response to be gaussian , but rather a non - linear function of @xmath36 , itself a linear function of the state .
the non - linearity and the noise characteristics of the observations follow from the choice of regression model made , e.g. , from the specific mapping between the signal and the response : choosing a gaussian distribution for the response with a mean equal to the signal yields the standard kalman filter . in this sense ,
our model is a generalization of the standard kalman filter .
a variant of the kalman filter known as the extended kalman filter deals with general non - linear observations , and has been applied to responses modeled through the exponential family ( @xcite , @xcite ) , yielding an algorithm that can be shown to be equivalent to ours . based on the assumptions above ,
we obtain the following factorization of the joint probability function of predictors , responses and parameters . @xmath72
we seek an algorithm to compute the mean and covariance matrix of the model parameters using all the observations we have at any given time .
we want this algorithm to be online , i.e. , to perform a relatively simple update to the previous mean and covariance estimates when a new observation arrives , without having to re - analyze previous observations .
let @xmath73 and @xmath74 denote the mean and covariance matrix of @xmath75 first , we initialize the mean and covariance of @xmath76 to the known values @xmath77 and @xmath78 .
we proceed by induction : we assume we know that @xmath79 has parameter mean and covariance matrix @xmath80 and @xmath81 , and use them and the new observation to compute @xmath82 and @xmath83 through a two stepped process suggested by the following simple relation : @xmath84 @xmath85 equation [ eq : bayes ] relates @xmath86 to its @xmath87 which predicts @xmath88 based on all previous information up to but ignoring the observation at time @xmath89 , and the log - likelihood of the latest observation @xmath90 .
we compute the mean and covariance of the prior @xmath91 via @xmath92 this equation is exact and does not require assuming any functional form for @xmath93 it follows fairly directly from equation [ eq : pardynamics ] , e.g. , see appendix [ sec : meanvarpredictionapp ] for a derivation of these and other equations in this section . when the parameter dynamics are non - linear , the mean and covariance of @xmath91 can be approximated through expressions identical to equations [ eq : meanpred ] and [ eq : varpred ] by suitably re - defining the matrices that appear in them , e.g. , by linearizing the parameter dynamics around @xmath94 or via numerical simulation as in the so - called unscented kalman filter ( @xcite , @xcite and @xcite ) . note
that when @xmath64 is the identity matrix , equation [ eq : varpred ] shows that the variance of the parameter estimates always increases in the prediction step , unless @xmath95 in addition , because the system input @xmath62 is deterministic , it does not contribute to the covariance matrix . when the predictor @xmath96 becomes known
, we can use equations [ eq : meanpred ] and [ eq : varpred ] to determine the mean @xmath97 and covariance matrix @xmath98 of the signal @xmath99 given @xmath100 and @xmath96 : @xmath101
lastly , the covariance matrix between the signal and the parameters is given by @xmath102 .
the latter follows from the fact that signal is a linear function of the parameters , with @xmath103 as the weights .
now we update the estimates from the prediction step to incorporate the new observation , obtaining the mean @xmath73 and covariance @xmath74 of the posterior @xmath75 we first compute the following matrices @xmath104^{-1 } + \mathbf{\omega_t } , \text { and } \label{eq : qmatrix } \mathbf{a_t } = & \mathbf{r_tx_tq_t^{-1}}.\end{aligned}\ ] ] here @xmath105 is the @xmath12-by-@xmath12 hessian matrix of the log - likelihood , evaluated at the predicted value of the signal @xmath106 as we will see , in many models of interest , this matrix is the negative of the variance of @xmath107 evaluated at the predicted signal value @xmath106 the matrix @xmath108 then grows with the expected variance of the predicted signal response @xmath98 , but decreases when the expected variance of the response increases .
we then compute the covariance @xmath74 via : @xmath109 computing the inverse of @xmath108 starting from equation [ eq : qmatrix ] can often be numerically unstable , e.g. , because the determinant of @xmath105 can be very small in magnitude .
a more robust way to compute @xmath110 is via @xmath111.\end{aligned}\ ] ] this expression follows directly from equations [ eq : varupdate2 ] and [ eq : qmatrix ] after applying the kailath variant of the woodbury identity ( e.g. , see @xcite ) .
we finally compute the mean of the parameters by : @xmath112 our main algorithm proceeds by executing the prediction and estimation steps for each arriving observation , namely evaluating equations [ eq : meanpred ] , [ eq : varpred ] , [ eq : varupdate2 ] and [ eq : meanupdate ] with every new observation .
equations [ eq : meanupdate ] and [ eq : varupdate2 ] are approximate , and follow from ( 1 ) a second - order taylor expansion of @xmath90 around @xmath113 , and ( 2 ) assuming the prior @xmath91 is gaussian with mean and covariance given by @xmath113 and @xmath114 .
a sketch of the argument is described in section [ sec : derivation ] . because the two assumptions we make are exact in the case of linear regression with a gaussian prior for @xmath115 equations [ eq : meanupdate ] and [ eq : varupdate2 ] are exact in that case and correspond to the standard kalman filter equations .
west _ et al . _
( @xcite ) make the different approximation that the prior @xmath79 is conjugate to the likelihood @xmath116 obtaining a slightly different algorithm that has only been developed for the univariate response scenario .
many regression models involve a scalar signal @xmath117 and a scalar response @xmath118 where the predictor is now simply a vector @xmath119 .
this is the situation for the most commonly encountered regression models , such as linear , logistic , poisson or exponential .
the matrices @xmath120 @xmath121 and @xmath108 then also become scalars , so the update equations [ eq : meanupdate ] and [ eq : varupdate2 ] simplify to @xmath122 where the predicted signal is @xmath123 .
the result is very appealing because no matrix inverses need to be computed .
here we consider models where the response is in the exponential family of equation [ eq : nef ] , and where the natural parameter @xmath34 is related to the signal via equations [ eq : glm1 ] and [ eq : glm2 ] .
the gradient in these models can be shown to be given by @xmath124 so the gradient is always proportional to the error @xmath125 , the difference between the response and its mean according to the model at the given signal .
the covariance matrix @xmath126 is in general a function of the signal @xmath127 but we drop that dependence in our notation below to reduce clutter . the hessian @xmath128 is then obtained by differentiating equation [ eq : gradientglm ] with respect to the signal once more , resulting in @xmath129 = \frac{\partial}{\partial \mathbf{\lambda_t}}\biggr [ \frac{\partial \mathbf{h(\lambda_t})'}{\partial \mathbf{\lambda_t } } \mathbf{\sigma_{y_t}^{-1 } } \biggr ] \big(\mathbf{y_t}-\mathbf{h(\lambda_t)}\big ) \nonumber \\ - \frac{\partial \mathbf{h(\lambda_t})'}{\partial \mathbf{\lambda_t } } \mathbf{\sigma_{y_t}^{-1 } } \frac{\partial \mathbf{h(\lambda_t})}{\partial \mathbf{\lambda_t}}. \label{eq : hessianglm } \end{aligned}\ ] ] evaluating equations [ eq : gradientglm ] and [ eq : hessianglm ] for a given choice of the likelihood and link function ( which determines the response function @xmath130 ) at the predicted signal @xmath131 , and plugging the resulting expressions into equations [ eq : varupdate2 ] and [ eq : meanupdate ] completes the algorithm . when the canonical link is used , the natural parameter is equal to the signal , so @xmath132 , and @xmath133 .
equations [ eq : gradientglm ] and [ eq : hessianglm ] simplify to @xmath134 so the gradient is proportional to the error , as before , and the hessian is proportional to the negative of the covariance of @xmath135 evaluating the gradient and hessian above at the predicted signal @xmath97 , and plugging in the resulting expressions into equations [ eq : varupdate2 ] and [ eq : meanupdate ] yields the update equations @xmath136 \mathbf{x_t'r_t } \label{eq : meanupdatecanonicalnefvar2 } \\ \mathbf{m_t } = & \mathbf{a_t}+ \mathbf{c_tx_t } \mathbf{\phi_t^{-1 } } \big(\mathbf{y_t}-\mathbf{h(f_t)}\big ) , \text { where } \label{eq : meanupdatecanonicalnef}\\ \mathbf{e_t}= & \mathbf{\phi_t^{-1 } \sigma_{y_t}(f_t ) \phi_t^{-1}}.\end{aligned}\ ] ] equation [ eq : meanupdatecanonicalnefvar2 ] is the numerically stable analog of equation [ eq : varupdate2 ] that avoids inverses of potentially close - to - singular matrices . * multivariate linear regression * is one of many examples that falls in this class of models .
there , @xmath137 and @xmath107 can be shown to be in the natural exponential family ( e.g. , see equation [ eq : gaussiannef ] in the appendix ) , with @xmath138 and @xmath139 and covariance matrix equal to @xmath140 which in this case is independent of the signal .
so the equations above yield the standard kalman filter equations .
@xmath141 other distributions in the natural exponential family , e.g. , the multinomial , have variances that are a function of the signal linear regression is the exception . *
the univariate signal case * covers the majority of applications encountered in practice . here
the signal @xmath142 , the response @xmath143 , and the nuisance parameter @xmath144 are all scalars , and the predictor @xmath119 is a vector , so the update equations become : @xmath145 with @xmath146 being the variance of the response evaluated at the predicted signal @xmath147 . equation [ eq : varupdatecanonicalnefunivariate ] shows that the effect of the new observation is to reduce the covariance of the parameters by an amount proportional to @xmath148 , and a gain that gets smaller when there is more variance in the predicted signal , as captured by @xmath149 and larger when the response is expected to have a higher variance @xmath146 .
many common regression models fall in this category .
* univariate linear regression * with @xmath150 and @xmath151 this is already in natural exponential form with @xmath152 , and with @xmath142 playing the role of the natural parameter ( i.e. , the canonical link was used to map the mean of the response to the signal ) . substituting @xmath153 and @xmath154 in equations [ eq : varupdatecanonicalnefunivariate ] and [ eq : meanupdatecanonicalnefunivariate ]
yields the univariate kalman filter .
@xmath155 in * poisson regression *
@xmath143 is a positive integer that follows a poisson distribution with mean @xmath156 .
the likelihood is @xmath157 which is again in natural exponential form with @xmath142 as the natural parameter , and with @xmath158 .
the variance of a poisson random variable is equal to its mean , so equations [ eq : varupdatecanonicalnefunivariate ] and [ eq : meanupdatecanonicalnefunivariate ] become @xmath159 in * exponential regression * @xmath143 is a non - negative real number that follows an exponential distribution with mean @xmath160 , so @xmath161 with mean @xmath162 and variance @xmath163 .
note that here @xmath164 , so unlike other models here , the update in the mean is negatively proportional to the error , namely : @xmath165 in * logistic regression * the response is a bernoulli random variable .
it takes on the value 1 with probability @xmath166 and the value 0 with probability @xmath167 .
so @xmath168 is the response function , and its inverse @xmath169 is the link function .
the likelihood becomes @xmath170 this is again in the natural exponential family with @xmath158 , and variance @xmath171 .
the last equation above implies that @xmath172 is indeed the canonical link .
so equations [ eq : varupdatecanonicalnefunivariate ] and [ eq : meanupdatecanonicalnefunivariate ] become @xmath173
we start from equation [ eq : bayes ] , and view the likelihood as a function of the model parameters , namely @xmath174 .
we then approximate @xmath175 about @xmath176 via the second - order taylor expansion : @xmath177 because the signal is given by @xmath178 we have that @xmath179 we also make the second approximation that @xmath180 .
this approximation is what is needed to make the mathematics below work out , but could be justified in that the gaussian distribution is the continuous distribution that has maximum entropy given a mean and covariance matrix , and these are the only known statistics of @xmath181 .
using the two approximations in equation [ eq : bayes ] results in @xmath182 being proportional to @xmath183 where @xmath184 equation [ eq : derivation2b ] follows from completing squares , e.g. , see appendix [ sec : completingsquares ] , and the proportional sign indicates that terms independent of @xmath185 were dropped .
the result shows that under our approximations , @xmath186 to finish the argument , we substitute the expressions in equation [ eq : derivation1 ] into equation [ eq : derivation3 ] , and apply the woodbury matrix inversion formula to the expression for @xmath74 in equation [ eq : derivation3 ] to finally get the update equations [ eq : varupdate2 ] and [ eq : meanupdate ] .
the models we discuss here can be and have been applied to a wide range of situations to model , analyze and forecast univariate and multivariate time series .
e.g. , see chapter 14 in @xcite or @xcite for a range of examples . here
we apply the models discussed to the contextual multi - armed bandits scenario , where so far only univariate time series modeled through a linear or logistic regression have been considered . in the latter case ,
the only treatment known to us approximates the covariance matrix as diagonal .
the models we have discussed enable explore / exploit algorithms for contextual multi - armed bandit scenarios where the reward depends on a multivariate response vector distributed according to the exponential family , and where the true parameters of the different arms are dynamic .
we hope this broadens the situations where contextual multi - armed bandit approaches can be helpful .
the standard setup involves a player interacting with a slot machine with @xmath187 arms over multiple rounds .
every time an arm is played a reward gets generated .
different plays of the same arm generate different rewards , i.e. , the reward is a random variable .
different arms have different and unknown reward distributions , which are a function of an observed context . at every time step
, the player must use the observed context for each arm and all the history of the game to decide which arm to pull and then collect the reward .
we seek algorithms that the player can use to decide what arm to play at every round in order to maximize the sum of the rewards received .
these algorithms build statistical models to predict the reward for each arm based on the context , and decide how to balance exploring arms about which little is known with the exploitation of arms that have been explored enough to be predictable .
the exploration / exploitation trade - off requires having a handle on the uncertainty of the predicted reward , so the models used need to predict at least the mean and variance of the reward for each arm .
real applications such as personalized news recommendations or digital advertising often have tight temporal and computational constraints per round , so the methods to update the statistical models with every outcome need to be online .
popular and useful model choices describe the univariate reward for each arm as a linear function of the context plus gaussian noise ( i.e. , through a linear regression , e.g. , see @xcite ) , or through a logistic regression ( @xcite ) . in the latter case ,
the algorithm that updates the model based on new observations uses a diagonal approximation of the parameter covariance matrix . in all these references , model parameters are assumed static ( although their estimates change with every observation ) . in the non - contextual multi - armed bandit problem ,
recent efforts have tried to generalize the distributions for the rewards to the univariate exponential family @xcite , and as far as we know this is the first treatment for the contextual case .
we consider the following scenario .
the parameters of all arms are compiled in the single parameter vector @xmath185 that , unlike other settings , is allowed to change over time according to equation [ eq : pardynamics ] .
some entries in @xmath185 correspond to parameters for a single arm , and others are parameters shared across multiple or all arms .
we describe the model parameters via @xmath188 where @xmath100 is the history of contexts and responses seen up to and including round @xmath189 . at the start of round @xmath89
, we observe the context matrix @xmath190 for each arm @xmath191 , and combine this information with our knowledge of @xmath192 to decide which arm to play .
denote the arm played by @xmath193 , and its corresponding context matrix simply by @xmath194 to make it consistent with the notation in the rest of this paper . playing arm @xmath193 results in a response @xmath107 with a distribution in the ( possibly multivariate ) exponential family that depends on the context @xmath195 the relation between the response and the context
is given by the dynamic glm in section [ sec : dynglm ] , so the mean of @xmath107 is a function of the signal @xmath196 the response is used to update our estimates of the model parameters @xmath197 , according to the algorithm described in section [ sec : dynglm ] , to be used in round @xmath198 .
we assume the reward @xmath199 received in round @xmath89 is a known deterministic function of the response , e.g. , a linear combination of the entries in @xmath107 .
if we knew the actual model parameters , the optimal strategy to maximize the rewards collected throughout the game would be to play the arm @xmath200 with the highest average reward , i.e. , @xmath201.$ ] we define the regret @xmath202 - e[f(\mathbf{y_t})|\mathbf{\lambda_t}=\mathbf{x'_t}\mathbf{\theta_t}],$ ] i.e. , the difference between the means of the rewards of the optimal arm and the arm played given the context and the model parameters .
unlike the more standard contextual setup , ours allow for the explicit modeling of parameter dynamics .
it also broadens the choice of probability distribution to use for the response or reward to more naturally match the model choice to the nature and dimensionality of the reward data .
e.g. , we can use a poisson regression when the reward is a positive integer , or have a response with multiple entries each with a different distribution , use all response entries to update our parameter estimates , and then define the reward to be a single entry in the response .
a contextual multi - armed bandit algorithm uses the knowledge of the parameters at each round and the context to decide which arm to play .
the widely used upper confidence bound ( ucb ) approach constructs an upper bound on the reward for each arm using the mean and covariance of the parameter estimates at every round , and selects @xmath193 as the arm with the highest upper bound , e.g. , see @xcite .
another approach that has gained recent popularity ( @xcite , @xcite ) is the so - called thompson sampling introduced in @xcite , where arm @xmath191 is selected at round @xmath89 with a probability that it is optimal given the current distribution @xmath203 for the model parameters .
it is only recently that asymptotic bounds for its performance have been developed both for the contextual ( @xcite , for the linear regression case only ) and the non - contextual ( @xcite ) case .
the studies mentioned have found thompson sampling to perform at pair or better relative to other approaches , and to be more robust than ucb when there is a delay in observing the rewards .
thompson sampling is also very easy to implement . in one variant
we sample a parameter value @xmath204 from the distribution @xmath205 and let @xmath206.$ ] in another variant , rather than sampling the model parameters once for all arms from @xmath205 we generate independent samples from the same distribution , the sample for arm @xmath191 denoted by @xmath207 , and then let @xmath208.$ ] the latter approach is found in @xcite for the linear regression case to have a total regret that asymptotically scales with the number of model parameters rather than with its square as in the first variant , so we use the second variant in our simulations .
our goal here is to demonstrate how our online regression models work in the contextual bandits case when the observations are multivariate and not gaussian , and when the model parameters are allowed to be dynamic .
the goal is not to compare different contextual bandit algorithms , so we only focus on thompson sampling .
the model we simulate is inspired by the problem of optimizing the netflix sign - up experience .
each arm corresponds to a variant of the sign - up pages that a visitor experiences a combination of text displayed , supporting images , language chosen , etc .
the context corresponds to the visitor s type of device and/or browser , the day of week , time of day , country where the request originated , etc .
some of these predictors are continuous , such as the time of day , and others are categorical , such as the day of the week . the goal is maximizing signups by choosing the sign - up variant the is most likely to lead to a conversion given the context .
we also observe other related outcomes associated to each visitor , such as the time spent on the sign - up experience , and whether they provide their email before signing up .
we assume that these other observations are also related to the model parameters ( though possibly with different context vectors ) , and use them to improve our parameter estimates .
so our response is multivariate , even if the reward is based on a single entry of the response vector .
lastly , we want to let the model parameters drift over time , because we know that different aspects of the netflix product are relevant over time , e.g. , different videos in our streaming catalog will be the most compelling in a month than today .
denote the response by @xmath209',$ ] and the reward by @xmath210 we model @xmath211 through a logistic regression , with a probability of taking the value 1 of @xmath212 and variance @xmath213 .
the two other response entries do not affect the reward , but we use them to improve our estimates of the model parameters .
we model @xmath214 as a linear regression with mean @xmath215 and variance @xmath216 , and @xmath217 through another logistic regression , with mean @xmath218 and variance @xmath219 .
the signal is @xmath220'=\mathbf{x'_t}\mathbf{\theta_t}.$ ] we assume that the entries of the response are independent of each other conditioned on the signal , so the nuisance parameter matrix @xmath221 is diagonal and time - independent , with the vector @xmath222'$ ] as its diagonal , and the covariance matrix of the response @xmath223 is diagonal with the vector @xmath224'$ ] as its diagonal .
the context matrix @xmath225 for arm @xmath191 has one row for each model parameter entry and one column per response entry .
some rows correspond to parameters shared by all arms , and others to parameters corresponding to a single arm . to construct @xmath226
we simulate continuous and categorical predictors that we sample at every round .
we let @xmath227 play the role of the continuous predictors , and sample each column from a zero - mean gaussian with covariance @xmath228 .
the diagonal entries in @xmath228 are sampled independently from an exponential distribution with rate of 1 , and the off - diagonal entries all have a correlation of @xmath229 .
we let the categorical predictor @xmath230 be a sample from a uniform categorical distribution with @xmath231 entries , i.e. , all entries in @xmath232 are zero except for one that is set to 1 .
we also let @xmath233 be an indicator vector that specifies that arm @xmath191 is being evaluated .
it has @xmath234 entries that are all zero except for its @xmath191-th entry which is set to 1 .
letting @xmath235 be a column vector with @xmath236 entries , all set to 1 , we define the context matrix for arm @xmath191 as @xmath237 here @xmath238 denotes the kronecker product between two vectors or matrices .
the first @xmath234 rows of @xmath226 simply specify what arm is being evaluated , the next @xmath239 rows correspond to the continuous predictors , followed by @xmath231 rows for the categorical predictors .
the next @xmath240 rows @xmath241 are the interaction terms between the continuous predictors and the arm ( only rows corresponding to the arm @xmath191 are non - zero ) , and the last @xmath242 rows are the interaction terms between the categorical predictor and the arm chosen ( all these rows are zero except one that is set to 1 ) .
the number of rows and model parameters is then @xmath243 .
we let @xmath244 and @xmath245 .
note that without the interaction terms , the optimal arm would be independent of the context .
we set the model parameter dynamics to @xmath246 where @xmath247 @xmath248 has diagonal entries that are independent exponential random variables with rate @xmath249 , and a correlation coefficient of 0.2 for its off - diagonal entries .
we sample a different matrix @xmath248 at every round .
we assume the first visitor arrives at @xmath250 , and start the game by sampling @xmath251 from a zero - mean gaussian with diagonal covariance matrix .
the diagonal entries are independent samples from an exponential distribution with rate equal to 1 .
we initialize the mean and covariance estimates of @xmath251 as @xmath252 and @xmath253 , where @xmath254 is the identity matrix . at round @xmath89 , starting from the mean @xmath80 and covariance @xmath81 estimates of the parameters , we compute the mean @xmath113 and covariance @xmath114 of the parameters .
we then sample one value of the model parameters for each arm from the resulting prior distribution , and construct the context matrices @xmath226 for each arm .
we use the context matrices and the parameter samples to choose @xmath193 ( which defines @xmath96 ) based on thompson sampling , we play @xmath193 and observe the response to obtain the round s reward , and update the parameter estimates to obtain @xmath73 and covariance @xmath74 and start the next round . , but using @xmath255 rather than @xmath249 to increase the diffusion rate of the model parameters.,title="fig:",width=264,height=264 ] , but using @xmath255 rather than @xmath249 to increase the diffusion rate of the model parameters.,title="fig:",width=264,height=264 ] figure [ fig : onerealization ] shows the result of one simulation with 2000 rounds and 10 arms labeled a through j. the left plot shows the optimal arm ( that with the highest predicted reward @xmath256 based on the actual parameters @xmath88 ) in blue and the arm played , selected via thompson sampling and the parameter estimates , in orange .
it is evident from the spread of the orange dots across arms that , as expected , there was more exploration at the start of the game .
the spread of the blue dots shows that the interaction terms between the context and the arm result in different arms being optimal in different rounds .
the right plot shows the fraction of rounds when the optimal arm was not chosen through the first @xmath89 rounds for all values of @xmath89 in the simulation .
it drops under 0.4 from close to 1.0 at the start .
the blue line on the same figure shows the cumulative average regret rate per round , which is the sum of regrets per round divided by the number of rounds .
the regret per round is simply @xmath257 , both evaluated using the actual parameters @xmath258 the yellow line shows the cumulative random regret that would have resulted from choosing any arm with equal probability , independently of the model parameter estimates or the context .
we then repeated the full simulation 30 times and averaged the resulting timeseries across runs , for different scenarios with a different number of arms .
figure [ fig : avgregrets ] shows the results . as expected , the probability of error , the regret and the random regret all increase with a larger number of arms .
but the increased regret rate is quite mild , and continues to drop as more rounds are played .
the benefit of the contextual bandit algorithm relative to uniformly at random choosing an arm is the difference between the random regret rate and the regret rate , and it increases nicely as the number of arms increases .
we expect our approach to fall apart when the parameters drift so quickly over time that the information in the observations is not enough to keep the covariance of the model parameters from growing .
we explored this by increasing the parameter diffusion rate by changing @xmath259 from @xmath260 to @xmath261 the results are shown in figure [ fig : avgregrets3 ] : although all metrics worsen , the regret rate still decreases nicely over time despite the large parameter fluctuations over time .
we described a framework to easily obtain online algorithms that approximately estimate the mean and covariance matrix of the model parameters for a wide range of multivariate regression models where the model parameters change over time .
although our derivation is novel , these algorithms have been well known within a subset of the time - series community for at least a decade , but to the best of the author s knowledge , are not well known within the broader machine learning and statistical community , where we think these tools can be helpful .
we also propose using the algorithms in the contextual multi - armed bandit problem , where the approach here allows for dynamic parameters and a wider range of reward distributions .
the methods we discuss here correspond to the so - called filtering problem in the kalman filter and related literature .
there are other related algorithms that solve the so - called smoothing problem , i.e. , that estimate the parameters at any point in the past using all the observations .
the latter have been useful for time - series analysis , but seem less obviously useful in machine learning applications ( though they are well known for the standard kalman filter , e.g. , see @xcite or @xcite ) , and so are not covered here .
also , in situations where the parameter dynamics are non - linear , or where higher moments of the parameter estimates are desired , there are good alternative simulation - based approaches , e.g. , that rely on ideas from importance sampling and particle filters , that may be better choices than the methods described here .
the best overviews of the full suite of methods that we know of are @xcite , @xcite and @xcite .
we thank devesh parekh and dave hubbard for the initial discussions that triggered this research , stephen boyd and george c. verghese for the suggestion to relate this to kalman filters , and to justin basilico , roelof van zwol , and vijay bharadwaj for useful feedback on this paper .
10 s. agrawal and n. goyal .
further optimal regret bounds for thompson sampling . in _ proceedings of the sixteenth international conference on artificial intelligence and statistics _ , pages 99107 , 2013 .
s. agrawal and n. goyal .
thompson sampling for contextual bandits with linear payoffs . in _ proceedings of the 30th international conference on machine learning ( icml-13 ) _ , pages 127135 , 2013 .
l. bottou .
large - scale machine learning with stochastic gradient descent . in _ proceedings of compstat2010
_ , pages 177186 .
springer , 2010 .
o. chapelle and l. li .
an empirical evaluation of thompson sampling . in _ advances in neural information processing systems _ , pages 22492257 , 2011 .
j. durbin and s. j. koopman .
time series analysis of non - gaussian observations based on state space models from both classical and bayesian perspectives . ,
62(1):356 , 2000 . j. durbin and s. j. koopman . .
number 38 .
oxford university press , 2012 .
l. fahrmeir .
posterior mode estimation by extended kalman filtering for multivariate dynamic generalized linear models .
, 87(418):501509 , 1992 . j. harrison and m. west .
springer , 1999 .
r. e. kalman . a new approach to linear filtering and prediction problems .
, 82(1):3545 , 1960 .
b. m. klein . .
phd thesis , citeseer , 2003 .
n. korda , e. kaufmann , and r. munos .
thompson sampling for 1-dimensional exponential family bandits . in _ advances in neural information processing systems _ ,
pages 14481456 , 2013 .
l. li , w. chu , j. langford , and r. e. schapire .
a contextual - bandit approach to personalized news article recommendation . in _ proceedings of the 19th international conference on world wide web _ ,
pages 661670 .
acm , 2010 .
t. minka . from hidden markov models to linear dynamical systems .
technical report , citeseer , 1999 .
c. n. morris . natural exponential families with quadratic variance functions . , pages 6580 , 1982 .
c. n. morris .
natural exponential families with quadratic variance functions : statistical theory . , pages 515529 , 1983 .
k. murphy .
filtering , smoothing and the junction tree algorithm . , 1999 .
j. a. nelder and r. baker .
generalized linear models . , 1972 .
k. b. petersen , m. s. pedersen , et al .
the matrix cookbook .
, 7:15 , 2008 .
d. simon . .
john wiley & sons , 2006 . w. r. thompson . on the likelihood that one unknown probability exceeds another in view of the evidence of two samples .
, pages 285294 , 1933 .
m. j. wainwright and m. i. jordan .
graphical models , exponential families , and variational inference . , 1(1 - 2):1305 , 2008 .
m. west , p. j. harrison , and h. s. migon .
dynamic generalized linear models and bayesian forecasting .
, 80(389):7383 , 1985 .
assuming that @xmath263 and using equation [ eq : pardynamics ] for the parameter dynamics , we have that @xmath264 = & e\big [ \mathbf{g_t\theta_{t-1 } } + \mathbf{b_t u_{t-1}}+\mathbf{\omega_t}|d_{t-1}\big ] \nonumber \\ = & \mathbf{g_t m_{t-1 } } + \mathbf{b_t u_{t-1 } } = \mathbf{a_t},\end{aligned}\ ] ] resulting in equation [ eq : meanpred ] .
here we used the assumption that @xmath265 = \mathbf{0}$ ] .
the covariance matrix @xmath114 of @xmath262 in equation [ eq : varpred ] is found as follows : @xmath266 \nonumber \\ = & e\big[\big(\mathbf{g_t}(\mathbf{\theta_{t-1 } } -\mathbf{m_{t-1}})+ \mathbf{\omega_t}\big)\big(\mathbf{g_t}(\mathbf{\theta_{t-1 } } -\mathbf{m_{t-1}})+ \mathbf{\omega_t}\big ) ' |d_{t-1}\big ] \nonumber \\ = & \mathbf{g_tc_{t-1}g_t'}+\mathbf{w_t}.\end{aligned}\ ] ] here , we used the assumption that the noise vector @xmath61 is uncorrelated with the parameter @xmath267 the mean and covariance @xmath268 $ ] of the signal are derived as follows : @xmath269 = & \mathbf{x_t'}e[\mathbf{\theta_t}|d_{t-1 } ] = \mathbf{x_t'a_t}=\mathbf{f_t}. \nonumber \\
\mathbf{\omega_t}= & e\big[\mathbf{x_t'}(\mathbf{\theta_t - a_t})(\mathbf{\theta_t - a_t } ) ' \mathbf{x_t } | d_{t-1},\mathbf{x_t } \big ] \nonumber \\ = & \mathbf{x_t'r_tx_t } .\end{aligned}\ ] ] so @xmath270 .
lastly , the covariance @xmath271 $ ] between the signal and the parameters at time @xmath89 given @xmath100 and the predictors @xmath96 is given by @xmath272 = \mathbf{x_t'r_t}.\end{aligned}\ ] ]
let @xmath25 be a random vector with @xmath10 entries distributed according to the exponential form @xmath273 where @xmath274 is a sufficient statistic for @xmath25 , @xmath34 is the natural parameter vector , and @xmath0 is a symmetric @xmath10-by-@xmath10 matrix and a nuisance parameter .
note that equation [ eq : nef ] is more restrictive , because it implicitly assumes that @xmath275 .
for example , if @xmath276 with known covariance matrix @xmath277 but unknown mean , we have that @xmath278 is @xmath279 so @xmath25 in in the exponential family with @xmath280 , @xmath46 is the natural parameter @xmath281 , and the sufficient statistic is @xmath282 . when the covariance matrix is unknown , the sufficient statistic expands to include @xmath283 and @xmath34 is a function of both @xmath46 and @xmath284 . if the covariance matrix is not known , then we can instead define the sufficient statistic to be @xmath285,$ ] where @xmath286 for any matrix @xmath1 is a column vector resulting from stacking all the columns in @xmath1 , and re - arrange equation [ eq : gaussiannef ] to have the natural exponential form in equation [ eq : ef ] , now with natural parameter @xmath34 being both a function of @xmath46 and @xmath140 with @xmath287 is proportional to : @xmath288}_{\mathbf{\eta}'}\underbrace{[\mathbf{y ' } \
vec(\mathbf{yy'})]'}_{\mathbf{t(y ) } } - \underbrace { \big(\frac{1}{2 } \mathbf{\mu}'\mathbf{\sigma^{-1 } } \mathbf{\mu } + \frac{1}{2}\log(|\sigma|)\big)}_{b(\mathbf{\eta } ) } , \label{eq : gaussiannef2}\end{aligned}\ ] ] so in this case @xmath289 and the natural parameter becomes a function of @xmath284 as well as of the mean @xmath46 . the function @xmath290 $ ] can be shown to equal @xmath291 .
@xmath292 the last equality follows because the integrand in the first term of the second equation is also a probability distribution in the natural exponential family with parameter @xmath293 , so the integral equals 1 . taking the first derivative of @xmath294 and evaluating it at @xmath295 yields @xmath296 = \frac{\partial b(\mathbf{\eta , \phi})}{\partial \mathbf{\eta}}. \label{eq : mgfmean}\end{aligned}\ ] ] similarly , the second derivative of @xmath294 at @xmath295 yields @xmath297 \mathbf{\phi^{-1 } } = \frac{\partial b(\mathbf{\eta , \phi})}{\partial \mathbf{\eta } } \biggr ( \frac{\partial b(\mathbf{\eta , \phi})}{\partial \mathbf{\eta}}\biggr ) ' + \frac{\partial^2 b(\mathbf{\eta , \phi})}{\partial \mathbf{\eta^2 } } , \text { so } \nonumber \\ \mathbf{\phi } \frac{\partial^2 b(\mathbf{\eta , \phi})}{\partial \mathbf{\eta^2 } } \mathbf{\phi } = e\biggr[\biggr(\mathbf{t(y)-e[\mathbf{t(y)}]}\biggr)\biggr(\mathbf{t(y)-e[\mathbf{t(y)}]}\biggr)'\biggr ] .
\label{eq : mgfvar}\end{aligned}\ ] ] setting @xmath298 above yields equations [ eq : nefmean ] and [ eq : nefvar ] .
the derivation of equation [ eq : derivation2 ] required turning the expression @xmath299 where @xmath300 @xmath301 ( which is symmetric and positive definite ) and @xmath302 into the so - called perfect square expression @xmath303 for @xmath304 .
we have @xmath305 so we need @xmath306 this implies that @xmath307 | we study the problem of estimating the parameters of a regression model from a set of observations , each consisting of a response and a predictor .
the response is assumed to be related to the predictor via a regression model of unknown parameters .
often , in such models the parameters to be estimated are assumed to be constant .
here we consider the more general scenario where the parameters are allowed to evolve over time , a more natural assumption for many applications .
we model these dynamics via a linear update equation with additive noise that is often used in a wide range of engineering applications , particularly in the well - known and widely used kalman filter ( where the system state it seeks to estimate maps to the parameter values here ) .
we derive an approximate algorithm to estimate both the mean and the variance of the parameter estimates in an online fashion for a generic regression model .
this algorithm turns out to be equivalent to the extended kalman filter .
we specialize our algorithm to the multivariate exponential family distribution to obtain a generalization of the generalized linear model ( glm ) . because the common regression models encountered in practice such as logistic , exponential and multinomial all have observations modeled through an exponential family distribution , our results are used to easily obtain algorithms for online mean and variance parameter estimation for all these regression models in the context of time - dependent parameters .
lastly , we propose to use these algorithms in the contextual multi - armed bandit scenario , where so far model parameters are assumed static and observations univariate and gaussian or bernoulli . both of these restrictions can be relaxed using the algorithms described here , which we combine with thompson sampling to show the resulting performance on a simulation . |
large anisotropic flow has been observed in ultra - relativistic nucleus - nucleus collisions at the relativistic heavy - ion collider ( rhic ) and the large hadron collider ( lhc ) @xcite .
anisotropic flow is an azimuthal ( @xmath0 ) asymmetry of the single - particle distribution @xcite : @xmath1 where @xmath2 is the ( complex ) anisotropic flow coefficient in the @xmath3th harmonic .
one usually uses the notation @xmath4 for the magnitude : @xmath5 .
anisotropic flow is understood as the hydrodynamic response to spatial deformation of the initial density profile .
this profile fluctuates event to event , which implies that the flow also fluctuates @xcite .
the recognition of the importance of flow fluctuations has led to a wealth of new flow observables , among which triangular flow @xcite and higher harmonics , as well as correlations between different fourier harmonics @xcite .
flow fluctuations provide a window @xcite into both the early stage dynamics and the transport properties of the quark - gluon plasma .
specifically , the magnitudes of higher - order harmonics ( @xmath6 to @xmath7 ) are increasingly sensitive to the shear viscosity to entropy density ratio @xcite .
the distributions of @xmath8 and @xmath6 carry detailed information about the initial density profile @xcite , while @xmath9 and higher harmonics are understood as superpositions of linear and nonlinear responses , through which they are correlated with lower - order harmonics @xcite .
ideally , one would like to measure the full probability distribution @xmath10 @xcite .
so far , only limited information has been obtained , concerning either the distribution of a single @xmath2 @xcite or specific angular correlations between different harmonics @xcite .
we propose to study the distribution @xmath10 via its moments in various harmonics @xcite , either single or mixed , and illustrate our point with realistic simulations using the ampt model @xcite . in sec .
[ s : method ] , we recall how moments can be measured simply with a single rapidity gap @xcite
. this procedure is less demanding in terms of detector acceptance than the one based on several rapidity windows separated pairwise by gaps @xcite , and can be used to study even four - plane correlators . in sec .
[ s : old ] , we list standard measures of flow fluctuations which have been used in the literature and express them in terms of moments . in sec . [
s : new ] , we introduce new observables which shed additional light on the origin of event - plane correlations .
for instance , a correlation between @xmath11 and @xmath9 has been observed , which increases with impact parameter @xcite .
this correlation is usually understood @xcite as an effect of the non - linear hydrodynamic response which creates a @xmath9 proportional to @xmath11 @xcite : the increase in the correlation is thus assumed to result from the increase of elliptic flow @xcite .
we show that this hypothesis can be tested directly by studying how the correlation between @xmath11 and @xmath9 is correlated with the magnitude of @xmath8 .
we also investigate in a similar way the origin of the three - plane correlation between @xmath8 , @xmath6 and @xmath12 @xcite .
the statistical properties of @xmath2 are contained in its moments , which are average values of products of @xmath2 , of the form @xmath13 where @xmath14 and @xmath15 are integers , and angular brackets denote an average value over events .
note that @xmath16 and @xmath17 .
azimuthal symmetry implies that the only nontrivial moments satisfy @xcite @xmath18 we now describe a simple procedure for measuring these moments , which applies to harmonics @xmath19 , i.e. , @xmath20 .
( we do not study here moments involving directed flow @xmath21 @xcite . )
we define in each collision the flow vector @xcite by @xmath22 where the sum runs over @xmath23 particles seen in a reference detector , and @xmath24 are their azimuthal angles . in eq .
( [ defqcomplex ] ) means that we choose to average over particles in each event @xcite , rather than summing @xcite or dividing by @xmath25 @xcite .
this choice is discussed at the end of sec .
[ s : old ] .
] one typically measures @xmath26 in two different parts of the detector ( `` subevents '' @xcite ) @xmath27 and @xmath28 , which are symmetric around midrapidity and separated by a gap in pseudorapidity ( i.e. , polar angle ) @xcite .
the moment ( [ defmoments ] ) is then given by @xmath29 which one can symmetrize over @xmath27 and @xmath28 to decrease the statistical error .
this configuration , with all factors of @xmath26 on one side and all factors of @xmath30 on the other side @xcite , suppresses nonflow correlations and self correlations as long as only harmonics @xmath19 are involved .
an alternative procedure , where self correlations are explicitly subtracted , is described in @xcite . in order to illustrate the validity of the method , we perform calculations using the ampt model @xcite .
ampt reproduces quite well lhc data for anisotropic flow ( @xmath31 to @xmath32 ) at all centralities @xcite .
the implementation adopted in this paper @xcite uses initial conditions from the hijing 2.0 model @xcite , which contains nontrivial event - by - event fluctuations .
flow in ampt is produced by elastic scatterings in the partonic phase .
in addition , the model contains resonance decays , and thus nontrivial nonflow effects . in the present work ,
subevent @xmath27 consists of all particles in the pseudorapidity range @xmath33 , and subevent @xmath28 is symmetric around mid - rapidity , so that there is an @xmath34 gap of @xmath35 between @xmath27 and @xmath28 @xcite .
the thumb rule for measuring moments is that smaller values of @xmath3 are easier to measure because @xmath4 decreases with @xmath3 for @xmath19 .
lower order moments , corresponding to smaller values of @xmath14 and @xmath15 , are also easier because higher - order moments are plagued with large variances , which entail large statistical errors .
, ( see eq . ( [ scaledmoments ] ) ) for @xmath36 , as a function of centrality , measured with the number of participant nucleons .
results are for ( a ) elliptic flow , @xmath37 , and ( b ) triangular flow , @xmath38 , in pb - pb collisions at @xmath39 tev .
open symbols represent ampt calculations and closed symbols are obtained from atlas data @xcite . ]
we first list observables which have been previously studied in the literature and explain how they can be measured using the method outlined in sec .
[ s : method ] .
fluctuations of @xmath4 have been studied using cumulants @xcite , which are linear combinations of even moments of the distribution of @xmath4 , that is , @xmath40 .
these moments are obtained by keeping only one value of @xmath3 and setting @xmath41 in eq .
( [ qmoments ] ) . figure [ fig : scal ] displays the scaled moments @xmath42 for @xmath36 as a function of centrality for @xmath37 and @xmath38 , obtained by using the subevent method of sec .
[ s : method ] .
the scaled moment @xmath43 thus defined is invariant if one multiplies @xmath4 by a constant , therefore it reflects the statistics of @xmath4 and should be essentially independent of the detector acceptance .
ampt calculations are in fair agreement with the atlas data @xcite , but tend to slightly overpredict @xmath43 , i.e. , overestimate flow fluctuations . if flow is solely created by fluctuations and if the statistics of these fluctuations is a 2-dimensional gaussian @xcite , then @xmath44 . as can be seen in fig .
[ fig : scal](b ) , @xmath45 for all centralities , as expected since @xmath46 is only from fluctuations in pb - pb collisions .
our simulation does not have enough statistics to detect this small non - gaussianity . ] similarly , as seen in fig .
[ fig : scal](a ) , @xmath47 is roughly equal to @xmath48 for central collisions where @xmath31 is mostly from gaussian fluctuations , but decreases for mid - central collisions , corresponding to the emergence of a mean elliptic flow in the reaction plane @xcite .
event - plane correlations @xcite can also be expressed in terms of moments which can be measured using the method outlined in sec .
[ s : method ] , as already discussed in ref .
. specifically , two - plane correlations are pearson correlation coefficients between moments .
the pearson correlation coefficient between two complex variables @xmath49 and @xmath50 whose average value is 0 , @xmath51 , is defined as @xmath52 @xmath53 in general , and @xmath54 if @xmath49 and @xmath50 are uncorrelated .
the correlation between the second and fourth harmonic planes , which is denoted by @xmath55 in ref .
@xcite , corresponds to @xmath56 , @xmath57 .
the largest source of uncertainty in this measurement is the denominator which involves @xmath58 , a measurement quadratic in the small harmonic @xmath9 . and @xmath59 as a function of centrality for @xmath60 , @xmath61 , @xmath62 , @xmath63 , in pb - pb collisions at @xmath39 tev calculated in the ampt model . ]
note that the scaled moments eq .
( [ scaledmoments ] ) are of the type @xmath64 , which is another measure of the correlation between @xmath49 and @xmath50 when @xmath65 and @xmath66 both differ from 0 .
this correlation measure equals unity if @xmath49 and @xmath50 are uncorrelated , and is larger than unity if there is a positive correlation .
it is in general easier to measure than the pearson correlation coefficient , because it does not involve the higher - order moments @xmath67 and @xmath68 .
the `` standard candles '' introduced in ref .
@xcite correspond to the case @xmath69 , @xmath70 , obtained by keeping only two harmonics , @xmath3 and @xmath71 , and setting @xmath72 in eq .
( [ qmoments ] ) .
these are correlations between the magnitudes @xmath4 and @xmath73 , which do not involve the angular correlation between event planes .
four of these correlations are displayed in fig .
[ fig : standardcandles ] .
the correlation between @xmath74 and @xmath75 is small and negative ( @xmath76 ) , as already seen in ampt calculations @xcite , while the correlation between the corresponding event planes is small and positive @xcite .
all other correlations are positive .
the correlation between @xmath77 and @xmath31 , and that between @xmath78 and @xmath46 , become smaller for more central collisions , which is likely due to the smaller non - linear contributions @xcite of @xmath77 and @xmath78 , respectively .
note that the observables in eq .
( [ scaledmoments ] ) and eq .
( [ pearson ] ) are defined in such a way that factors of @xmath79 in eq .
( [ defqcomplex ] ) cancel between the numerator and the denominator if @xmath23 is the same for all events . in general
, @xmath23 fluctuates , and the result depends on whether or not one includes a factor @xmath79 .
however , the centrality selection in experiments is typically done using the multiplicity in a reference detector ( see e.g. ref .
@xcite ) so that effects of multiplicity fluctuations are likely to be small in narrow centrality intervals .
the calculations in this paper are done with a @xmath79 normalization , but there is no strong argument for preferring one normalization over another .
) , ( [ corrfluct2 ] ) and ( [ corrfluct3 ] ) .
circles connected by solid lines correspond to the left - hand sides , and squares connected by dashed lines correspond to the right - hand sides of these equations . ]
we now introduce new correlation measures of the type @xmath64 in order to study how event - plane correlations are correlated with the magnitude of anisotropic flow .
we first consider the case @xmath80 and @xmath81 .
@xmath82 is obtained by setting @xmath83 and @xmath84 in eq .
( [ qmoments ] ) .
the correlation @xmath64 is displayed in fig . [
fig : all](a ) .
there is a significant positive correlation for all centralities , which becomes larger for central collisions . in hydrodynamics ,
the correlation between @xmath9 and @xmath11 originates from a non - linear response @xcite . in order to test this hypothesis
, we model @xmath9 as the sum of two terms : @xmath85 where the non - linear response coefficient @xmath86 is the same for all events in a centrality class .
this corresponds to the separation of @xmath9 into a linear part , created by fluctuations , and a non - linear part from @xmath8 @xcite .
we assume in addition that @xmath87 and @xmath8 are uncorrelated .
equation ( [ model ] ) then implies the following relation between moments : @xmath88 this equation relates a mixed correlation between @xmath9 and @xmath8 to the fluctuations of elliptic flow : the right - hand side is the scaled moment @xmath89 introduced in eq .
( [ scaledmoments ] ) .
the ampt simulations support eq .
( [ corrfluct1 ] ) for all centralities , as can be seen in fig .
[ fig : all](a ) . a straightforward generalization of eq .
( [ corrfluct1 ] ) is obtained using @xmath90 instead of @xmath74 : @xmath91 this correlation gives a higher weight to events with large elliptic flow .
equation ( [ corrfluct2 ] ) is also supported by ampt simulations , as shown in fig .
[ fig : all](b ) .
this discussion can be readily extended to the correlation between the 5th harmonic plane and the 2nd and 3rd harmonic planes .
we now set @xmath92 and @xmath81 or @xmath93 and write @xmath94 where @xmath95 is independent of @xmath8 and @xmath6 and @xmath96 is constant .
one thus obtains @xmath97 the numerators in the right - hand side are obtained using eq .
( [ qmoments ] ) with @xmath98 , @xmath99 ( first line ) and @xmath100 , @xmath101 ( second line ) . figs .
[ fig : all](c)-(d ) again show that these equalities are very well verified by ampt simulations .
one can also use the fact that @xmath31 and @xmath46 are weakly correlated , as seen in fig .
[ fig : standardcandles ] , which leads to the following simplified relations : @xmath102 these relations are also satisfied to a good approximation ( see figs . [
fig : all](c)-(d ) ) , thus showing that the correlators in eq .
( [ corrfluct3 ] ) are mostly driven by the fluctuations of @xmath31 and @xmath46 .
moments of the distribution of @xmath2 provide a complete set of multiparticle correlation observables , which can be used to probe the physics of flow fluctuations in unprecedented detail .
all these moments can be measured using just two subevents separated by a rapidity gap .
moments yield the full information on the multiparticle correlations without resorting to unfolding procedures @xcite or event - shape engineering @xcite .
they can be measured easily at lhc and even with detectors having smaller acceptance , and can be directly compared with theoretical calculations . in particular , scaled moments @xcite as studied in this paper are typically independent of the details of the acceptance , and reflect global fluctuations .
they can be used to further probe the physics of initial - state fluctuations and the hydrodynamic response of the quark - gluon plasma .
we have shown that the assumption that the correlation between @xmath9 and @xmath8 , and that between @xmath12 , @xmath8 and @xmath6 , are driven by nonlinear response , yields nontrivial relations between moments , eqs .
( [ corrfluct1 ] ) , ( [ corrfluct2 ] ) and ( [ corrfluct3 ] ) .
simulations within the ampt model have shown that these relations are very well satisfied .
it is important to test if experimental data confirm these predictions .
this work is funded by cefipra under project 4404 - 2 .
jyo thanks the mit lns for hospitality and acknowledges support by the european research council under the advanced investigator grant erc - ad-267258
. l. v. bravina , b. h. brusheim johansson , g. k. eyyubova , v. l. korotkikh , i. p. lokhtin , l. v. malinina , s. v. petrushanko and a. m. snigirev _ et al .
_ , phys .
c * 89 * , no . 2 , 024909 ( 2014 ) [ arxiv:1311.0747 [ hep - ph ] ] . | we present a complete set of multiparticle correlation observables for ultrarelativistic heavy - ion collisions . these include moments of the distribution of the anisotropic flow in a single harmonic , and also mixed moments , which contain the information on correlations between event planes of different harmonics .
we explain how all these moments can be measured using just two symmetric subevents separated by a rapidity gap .
this presents a multi - pronged probe of the physics of flow fluctuations .
for instance , it allows to test the hypothesis that event - plane correlations are generated by non - linear hydrodynamic response .
we illustrate the method with simulations of events in a multiphase transport ( ampt ) model . |
the two nearest galaxy clusters to us are virgo and fornax .
the nasa extra - galactic database ( ned ) gives the distance of m87 ( the giant elliptical at the heart of virgo ) as 16.8 mpc , while the similar galaxy at the heart of fornax ( ngc1399 ) lies at 17.9 mpc .
both clusters are at high galactic latitude ( @xmath9 , @xmath10 ) and so are ideal targets for extra - galactic studies at all wavelengths .
virgo and fornax were observed at optical wavelengths ( photographic ) in the 1980s leading to the most complete catalogues of cluster members ( binggeli et al .
1985 , ferguson 1989 ) .
fornax is somewhat less populous than virgo with just 350 members listed in the fornax cluster catalogue ( fcc , ferguson 1989 ) compared to about 2000 in the virgo cluster catalogue ( vcc , binggeli et al .
1985 ) .
the fornax cluster has a velocity dispersion of @xmath11 km s@xmath12 and a core radius of 0.7 deg ( 250 kpc ) , which gives a mass of @xmath13 m@xmath8 , about a factor of seven less than virgo ( jordan et al .
2007 and references therein ) and so fornax might also be described as a large group . like virgo
, fornax does show evidence of sub - structure indicating its current and continuing assembly ( drinkwater et al . 2001 ) though it is more regular in shape and probably more dynamically evolved than virgo .
although less massive than virgo , the central galaxy number density is about three times higher in fornax than in virgo and with its lower velocity dispersion we might expect more influential interactions of fornax galaxies with one another . to quantify this the relaxation time of fornax , a measure of the time scale for gravitational interactions ,
is about one third that of virgo ( @xmath14 years , boselli and gavazzi , 2006 ) .
however , this timescale is long compared to the crossing time which determines how long a mechanism like ram pressure stripping might be effective .
the crossing time of fornax and virgo are about equal ( @xmath15 years , boselli and gavazzi , 2006 ) , but the lower velocity dispersion of fornax makes the inter - galactic medium less influential for the same inter - galactic medium density .
we can quantify how effective ( @xmath16 ) ram pressure stripping might be using @xmath17 ( gunn and gott , 1972 ) where @xmath18 is the cluster velocity dispersion and @xmath19 is the gas density . from the x - ray data ( scharf et al .
2005 , paolillo et al .
2002 , schindler et al .
1999 ) the central gas density and velocity dispersion of virgo is about 4 and 2 times higher respectively than fornax making virgo potentially about 16 times more effective at ram pressure stripping .
so , because of these diverse cluster properties virgo and fornax galaxies may have quite different gas accretion / stripping histories .
our new observations of fornax cluster galaxies described below make use of the unique imaging qualities and sensitivity of the herschel space telescope at five wavelengths from 100 - 500@xmath0 m . with a 3.5 m mirror the spatial resolution of herschel at these wavelengths ranges from 7 - 36@xmath20enabling spatially resolved observations of many of the bright galaxies at the distance of fornax ( 10@xmath20@xmath21 kpc ) .
the herschel fornax cluster survey ( hefocs ) is an esa herschel space observatory ( pilbratt et al
. , 2010 ) open time project .
the project was awarded 31 hours of observing time in parallel mode using pacs ( poglitsch et al . , 2010 ) at 100 and 160@xmath0 m and spire
( griffin et al . , 2010 ) at 250 , 350 and 500@xmath0 m .
pre - herschel comparable surveys describing the far infrared properties of nearby bright galaxies in both cluster and field have been described by soifer et al .
( 1987 ) , doyon and joseph ( 1989 ) ( iras ) , tuffs et al , ( 2002 ) ( iso ) , draine et al .
( 2007 ) ( spitzer ) .
davies et al .
( 2012 ) have presented herschel observations of the bright galaxies in the virgo cluster . in this paper
we describe the first results from our fornax cluster survey concentrating on the properties of the bright galaxies that are detected at high signal - to - noise . in many ways
this paper is a continuation of the work we have previously done using the herschel virgo cluster survey ( hevics ) data , which is described in the following series of papers : * paper i * ( davies et al .
2010 ) that considered the properties of the bright galaxies in a single central 4@xmath224 sq deg hevics field , * paper ii * ( cortese et al .
2010 ) that describes the truncation of cluster galaxy dust discs , * paper iii * ( clemens et al . 2010 ) on the dust life - time in early - type galaxies , * paper iv * ( smith et al . 2010 ) considers the spiral galaxy dust surface density and temperature distribution , * paper v * ( grossi et al .
2010 ) on the properties of metal - poor star - forming dwarf galaxies , * paper vi * ( baes et al . 2010 ) looks at the lack of thermal emission from the elliptical galaxy m87 , * paper vii * , ( de looze et al .
2010 ) discusses the far - infrared detection of dwarf elliptical galaxies , * paper viii * ( davies et al .
2012 ) looks at the properties of the 78 brightest virgo galaxies , * paper ix * ( magrini et al .
2011 ) compares galactic metalicity and dust - to - gas ratio gradients , * paper x * ( corbelli et al .
2012 ) considers the relationship between cold dust and molecular gas , * paper xi * ( pappalardo et al .
2012 ) assesses the environmental effects on molecular gas and dust in spiral disks and * paper xii * ( auld et al .
2012 ) discusses the detection of far - infrared emission from all catalogued vcc galaxies . a further paper ( boselli et al . , 2010 ) discusses the spectral energy distributions of hevics galaxies together with others observed as part of the herschel reference survey ( hrs ) .
we have obtained @xmath23 sq deg of data from a single field centred on the central fornax cluster galaxy ngc1399 ( ra(j2000)=03:38:29.08 , dec(j2000)=-35:27:2.7 ) , fig .
1 . we use parallel mode and a fast scan rate of 60/sec ( as used for the hevics data ) over two@xmath22two orthogonal cross - linked scan directions ( 4 scans ) .
pacs data reduction was carried out with the standard pipeline for both the 100 and 160@xmath0 m channels up to level-1 .
the four scans were then combined and maps made using the _ scanamorphos _ map - maker ( roussel , 2012 ) .
we chose to use _ scanamorphos _ , instead of the phot - project mapper used in previous papers ( davies et al . , 2012 ) , as this avoids the necessity of high - pass filtering , which would remove the lower level extended emission . after combining orthogonal scans
the full width half maximum ( fwhm ) beam sizes are approximately 7and 13with pixel sizes of 2and 3for the 100 and 160 @xmath0 m channels respectively .
the spire data were also processed up to level-1 with a custom driven pipeline script adapted from the official pipeline ( _ pof5_pipeline.py _ , dated 8 jun 2010 ) as provided by the spire instrument control centre ( icc ) .
this _ jython _
script was run in the herschel interactive processing environment ( hipe - ott , 2010 ) .
our data reduction up to level-1 is very similar to the herschel common science system / standard product generation v5 with a calibration based on neptune data .
specific differences to the standard pipeline were that we used the _ sigmakappadeglitcher _ instead of the icc - default _
waveletdeglitcher_. furthermore , we did not run the default _ temperaturedriftcorrection _ and the residual , median baseline subtraction .
instead we use a custom method called brigade ( smith et al . in preparation ) to remove the temperature drift and bring all bolometers to the same level ( equivalent to baseline removal ) .
we have found this method improves the baseline subtraction significantly especially in cases where there are strong temperature variations during the observation .
scans were then combined to make our final maps using the naive mapper provided in the standard pipeline .
the fwhm of the spire beams are 18.2 , 24.5 , and 36.0with pixel sizes of 6 , 8 , and 12at 250 , 350 , and 500 @xmath0 m , respectively .
the beam areas used were 423 , 751 and 1587 sq .
arc sec respectively .
the final data products have a 1@xmath24 noise , determined from the whole of each image , of @xmath250.6 , 0.7 , 1.0 , 1.0 and 1.0 mjy pixel@xmath12 or 6.4 , 3.3 , 1.2 , 0.7 and 0.3 mj sr@xmath12 at 100 , 160 , 250 , 350 and 500 @xmath0 m respectively . as in davies et al .
( 2012 ) for the virgo cluster we obtain a bright far - infrared selected sample by carrying out our initial object selection at 500 @xmath0 m .
this is because it is the least explored part of the spectrum , it has the lowest resolution , and most galaxies will produce their lowest flux in this band , which should guarantee a detection in all five bands . to produce an objectively selected sample we used the automatic image detection algorithm sextractor ( bertin and arnouts , 1996 ) . to minimise background contamination by faint sources each object
was required to have more than 41 connected pixels at 1.5@xmath26 or above.m and so the two are equivalent . ]
the result is a 500 @xmath0 m flux density limit of @xmath250.1 jy for sources with a diameter larger than 1.4 .
each object was then checked for correspondence with a known fornax cluster galaxy ( fcc ) .
the final sample consists of 11 fornax cluster objects , which is just 4% of the 256 fcc galaxies in our field .
this compares with 78 galaxies detected in just the same way in the virgo cluster , which is 12% of the galaxies in this area listed in the vcc ( davies et al .
a more straight forward comparison is that 14 galaxies were detected in the central 16 sq deg virgo field centred on m87 , compared to 11 over the same area centred on ngc1399 .
discarded objects are in the background of fornax .
in the standard way the data were initially smoothed and re - gridded to the 500 @xmath0 m resolution and pixel scale .
elliptical apertures were chosen by eye using the 500@xmath0 m data , with the sky defined by a concentric annulus .
these same annuli were then used on the smoothed and re - gridded data at other wavelengths . as a test we compared this method with carrying out the aperture photometry on the original unsmoothed 100@xmath0 m data , our criteria being correspondence with the flux density values at 100@xmath0 m from iras .
these tests showed that we obtained a much better result by using the original resolution data , rather than the smoothed and re - gridded data .
so , independent measurements were carried out on the maps at the original resolution , defining by eye for each galaxy an elliptical aperture and a nearby sky aperture that avoided pixels contaminated by bright background sources .
the flux densities quoted in table 1 come from aperture photometry using this unsmoothed and original resolution data .
m and the planck ( red ) flux densities with the spire 350@xmath0 m .
linear least squares fit parameters are given top left . the black dashed line is the one to one relationship . ] [ cols="<,^,^,^,^,^,^,^,^,^ " , ] where available ( 8 out of 11 galaxies ) , we have taken the 21 cm spectral line flux integral ( @xmath27 , jy km s@xmath12 ) for each galaxy from ned and used the distances given in table 1 to obtain atomic hydrogen gas masses ( @xmath28 , m@xmath8 ) .
the masses have been obtained using the standard formula : @xmath29 atomic hydrogen gas masses are given in table 3 along with dust and stellar mass . derived from co observations
there are 4 detections and 3 upper limits on the molecular hydrogen masses of these galaxies ( horellou et al .
nc1365 , ngc1386 , ngc1436 and eso358-g063 have log(@xmath30 ) equal to 10.3 , 8.3 , 7.7 and 8.6 respectively .
there is almost a factor of 2 more molecular hydrogen in ngc1365 than atomic and ngc1386 is surprisingly detected in co , but not hi .
ngc1351a , ngc1350 and ngc1387 all have upper limits on their molecular gas masses that are in each case more than a factor of 10 less than their atomic hydrogen masses
. we have not incorporated this molecular gas data into table 3 or discussed it further so that this paper is directly comparable with the equivalent paper on virgo , where we have not included molecular gas masses ( davies et al . 2012 ) . in the above we
have taken @xmath31 @xmath32 ( k km / s)@xmath12 . by simply dividing the total mass in each component by the volume sampled ( @xmath2512.9(7.2 ) mpc@xmath33 )
we can calculate the cluster mass densities . provided the luminosity distributions in each far - infrared band are peaked the total cluster dust mass should be well constrained by our sample galaxies unless there is a significant dust component in the inter - galactic medium . from the dust in our sample galaxies
we derive the densities given in table 4 .
values for virgo have been taken from davies et al .
these values can be compared with a recent determination of the local dust mass density for galaxies in all environments from dunne et al .
this shows that fornax is over dense in dust by about a factor of 120(196 ) while virgo is over dense by about a factor of 39(126 ) . if the fornax hi mass function is peaked , similar to virgo ( davies et al .
2004 , taylor 2010 ) , the hi gas in these galaxies should provide a good estimate of the cluster total . using the eight hi masses available ( table 3 ) we derive the hi mass density shown in table 4 .
the value for virgo is taken from davies et al .
recently , davies et al .
( 2011 ) measured a local field hi mass density of @xmath34 m@xmath8 mpc@xmath35 ( see also martin et al .
thus the fornax cluster is over dense in hi , when compared to the field , by a factor of 18(29 ) and virgo by 6(18 ) . as a check of our assumption that our sample galaxies contain the majority of hi in the volume we have looked at the hipass data for fornax as described in waugh et al .
there are three galaxies in our herschel field that have hi detections by waugh et al . , but are not in our sample ( a visual inspection shows that they are all detected in the far - infrared , but not at the required flux level ) .
the three hi detections are eso358-g015 , eso358-g051 and eso358-g060 and using their distances as listed in ned they amount to just 9% of the total atomic hydrogen in our sample galaxies .
comparing stellar mass functions is not quite so straight forward because of uncertainties in the faint end slope of the cluster luminosity function i.e the numbers of faint dwarf galaxies ( see sabatini et al .
, 2003 , and references therein ) . for the stars in our sample galaxies we obtain the densities given in table 4 .
the virgo value is again taken from davies et al .
baldry et al .
( 2008 ) recently derived the local galaxy stellar mass function and from it the field stellar mass density ( table 4 ) .
thus the fornax cluster is over dense in stars by 120(195 ) compared to values for virgo of 34(129 ) . where required in the above derivations
we have used h@xmath36=72 km s@xmath12 , @xmath37 , @xmath38 . the larger stellar mass density
we calculate for fornax compared to virgo is again roughly consistent with the factor of 2.5 higher number density of galaxies derived by ferguson ( 1989 ) - fornax is a higher over density than virgo .
interestingly , both virgo and fornax represent about equal over densities of dust and stars , but the atomic gas is relatively much more depleted ( table 4 ) .
this presumably indicates that atomic gas has been removed from these galaxies without the loss of consummate amounts of dust , though cortese et al .
( 2010 ) show , by the clear truncation of virgo cluster dust discs , that some dust must be lost from the outer regions of galaxies in the cluster environment .
we can make a simple calculation to see the effect of ram pressure on both the gas and the dust .
the force per unit area a particle in the disc of a galaxy experiences is the difference between the ram pressure and the restoring force per unit area due to the discs surface mass density ( gunn and gott , 1972 ) . in this case
the acceleration a particle undergoes depends on its cross - sectional area ( @xmath39 ) and its mass ( @xmath40 ) and for constant acceleration the timescale for removal is just proportional to @xmath41 .
using typical values for the masses ( @xmath42 , @xmath43 ) and sizes ( @xmath44 , @xmath45 ) of gas and dust particles leads a timescale for dust removal about a factor of 100 - 1000 longer than gas removal ( this ignores viscous drag between the gas and dust , see davies et al .
evidence from virgo shows that atomic and molecular gas are also affected very differently by the cluster environment .
there is no evidence for a h@xmath46 deficiency in virgo but there clearly is an atomic gas deficiency ( gavazzi et al .
2008 , young et al .
so , a complementary explanation for the non - removal of dust is that it is primarily associated with dense h@xmath46 reservoirs that are very resistant to the ram pressure stripping process . using the davies et al .
( 2010 ) and dunne et al .
( 2011 ) values for the global hi and dust mass densities gives a local field atomic gas to dust ratio of 363 ( draine et al .
, 2007 ) , but our calculation neglects the molecular gas included in this value . ] , while the same ratio in the fornax and virgo clusters is 52(53 ) and 58(50 ) respectively . from the above it
seems that these low values are entirely due to the loss of atomic gas . in each case
this amounts to about six times as much hi lost from these galaxies as is presently contained within them ( @xmath47 m@xmath8 and @xmath48 m@xmath8 respectively for virgo and fornax ) .
it seems that irrespective of the different environments prevalent in these two clusters the atomic gas loss mechanisms are equally effective .
hi complexes within the virgo cluster , but external to previously identified galaxies , have been found by davies et al .
( 2004 ) , oosterloo and van gorkom ( 2005 ) and kent et al .
( 2007 ) , but they only represent a small fraction of the cluster atomic gas .
the brightest virgo cluster galaxy at these wavelengths is ngc4254 ( @xmath49 jy , @xmath50 mpc , davies et al . 2012 ) , but this is totally outshone by the fornax cluster galaxy ngc1365 ( @xmath51 jy , @xmath52 mpc ) .
being so bright ngc1365 is also detected by planck in 4 bands at 350 , 550 , 850 and 1382@xmath0 m . in fig 4
we show the pacs , spire and planck data along with the best fitting modified blackbody taken from fig 3 .
the planck data are those derived by fitting a 2d gaussian to the data ( gauflux , which should be more appropriate for extended sources ) and color corrected using the herschel sed and the tables in the explanatory notes of the planck early release compact source catalogue .
the planck data fit remarkably well the modified blackbody all the way out to 1382@xmath0 m .
there is no significant sign of any excess emission above that of the modified blackbody , with the point at 1382@xmath0 m being the most discrepant .
the band centred on 1382@xmath0 m contains the co(2 - 1 ) line and this might explain the slight excess . according to sandqvist et al .
( 1995 ) the total co(1 - 0 ) luminosity of ngc1365 is 5200 jy km / s .
for the planck bandwidth ( 33% of the central frequency ) and using the measured co(2 - 1)/co(1 - 0)=0.8 ratio we find that the co(2 - 1 ) line contributes 0.045 jy ( 7% of the total ) to the 1382@xmath0 m flux density .
if this is subtracted from the color corrected data , than the 1382@xmath0 m point sits nicely on top of the modified blackbody curve . over these
longer , planck wavelengths , we might also expect an additional contribution to the observed flux density from free - free emission ( synchrotron emission only starts to become important at wavelengths greater than @xmath251 cm , condon , 1992 ) . combining equations 23 and 26 from condon ( 1992 )
we get : + @xmath53 where @xmath54 is the free - free flux density in jy at wavelength @xmath55 in @xmath0 m , @xmath56 is the distance in mpc and @xmath57 the far - infrared luminosity in solar units . for ngc1365 at a distance of @xmath58 mpc and @xmath59 @xmath60 the predicted flux densities at 850 and 1382@xmath0 m are 0.048 and 0.050 jy respectively .
these values are again much smaller than the extrapolation of the modified blackbody curve and so free - free emission is predicted to be only a small fraction of the emission at these wavelengths .
ngc1365 is a powerful seyfert ( type 1.8 ) galaxy ( veron - cetty and veron 2006 ) , but no sign of the agn is seen in its global far - infrared / sub - mm spectrum .
previous observations of some more passive galaxies has revealed an excess at wavelengths beyond 350@xmath0 m , particularly the milky way ( leitch et al . 1997 ) and star forming dwarf galaxies ( grossi et al . 2010 ) .
this emission has been associated with an additional spinning dust component ( draine and lazarian , 1998a , 1998b ) , but it is not observed here . with a b band absolute magnitude of -20.7 ngc1399
is just about a typical ( @xmath61 ) galaxy in terms of its optical luminosity .
but , being an elliptical galaxy ( e1 ) lying at the very heart of the cluster we might expect that ngc1399 s inter - stellar medium is somewhat depleted compared to a galaxy like the milky way .
no atomic hydrogen has been detected in ngc1399 , though there have been previous detections in the far - infrared possibly associated with dust emission .
there is an iras detection at 0.3 jy ( ned ) which would be a marginal detection in our data dependent on the size of the source - but we detect nothing .
there is a _ spitzer _ ( mips ) detection at 160@xmath0 m of 0.026 jy ( temi et al .
2009 ) , but this is below our detection limit . for the similar , though slightly brighter in the optical m87 , our observations gave detections in all the pacs / spire bands ( davies et al .
2012 ) , though this was almost exclusively synchrotron emission and not thermal emission from dust ( baes et al .
2010 ) . for synchrotron emission from ngc1399
we expect the highest flux density at the longest wavelengths , consistent with our detection at 500@xmath0 m and our non - detection at 100@xmath0 m .
for the same spectral slope as m87 we would expect a 100@xmath0 m flux density of about 0.1 jy , below our detection threshold and rather at odds with the iras 100@xmath0 m detection ( which is too high ) and similarly at odds with the mips measurement ( which would be too low ) .
there are radio synthesis observations ( 44resolution ) of ngc1399 at 35 cm that clearly show two opposing radio jets emanating from the nucleus ( jones and mcadam , 1992 ) . in fig .
5 we show the 500@xmath0 m image of ngc1399 with the position and extent of the 35 cm radio jet indicated by the red arrow .
the 500@xmath0 m emission appears more extended than the jet and also the jet seems to be off - set from it , but roughly at the same position angle .
the optical centre of the galaxy ( ned ) corresponds with the centre of the jet . for m87
the far - infrared emission clearly traces the radio jet and so it is a bit of a puzzle as to why ngc1399 is different .
if the emission is actually due to dust , and this is some form of dust lane , then it is strange that it is not detected more strongly in the other spire bands .
it also extends over some 3 arc min ( 16 kpc ) - this is large for a dust feature in an elliptical galaxy - ngc1399 does not appear in the catalogue of dusty ellipticals ( ebneter and balick , 1985 ) .
there is also the possibility that this is some chance alignment of background sources .
the region shown in fig . 5 is some 20 arc min across , specifically made this large to illustrate the brightness and number density of sources across the field .
there are background sources that could conspire to produce the emission we are associating with ngc1399 , and so at the moment the origin of this 500@xmath0 m emission is not clear .
m image ( radius @xmath62 arc min ) centred on the position of ngc1399 .
extended emission is seen at approximately the position of ngc1399 .
the red arrow shows approximately the position , extent and orientation of the radio jet .
the centre of the galaxy when measured in the optical and uv is indicated by the white arrow . ]
in this paper we have described the herschel fornax cluster survey ( hefocs ) and far - infrared observations of 11 bright cluster galaxies .
we carry out aperture photometry and show that our measurements are consistent with previous data where available . for 10 of the 11 galaxies we fit spectral energy distribution and derive dust masses and temperatures .
we combine dust masses with star and gas masses from the literature to assess the global properties of the galaxies in our sample .
these global properties are similar to those detected by us in the virgo cluster ( hevics , davies et al .
we have then compared fornax and virgo with each other and the field .
generally fornax is more dense in stars , gas and dust by a factor of 3 - 4 relative to virgo . when compared to the lower density local field of galaxies , fornax is overdense by a factor of about 120 in stars and dust , but only a factor of 18 in gas .
values for virgo are about 34 and 39 for stars and dust , reducing to 6 for gas
. both the fornax and virgo clusters are relatively depleted in gas to the same extent - gas loss does not seem to be dependent on these two different galactic environments .
we specifically discuss two of the sample galaxies in more detail . ngc1365 because it is the brightest far - infrared galaxy in virgo and fornax combined .
ngc1399 because it is only detected at the longest wavelength of 500@xmath0 m .
* acknowledgements * + the herschel spacecraft was designed , built , tested , and launched under a contract to esa managed by the herschel / planck project team by an industrial consortium under the overall responsibility of the prime contractor thales alenia space ( cannes ) , and including astrium ( friedrichshafen ) responsible for the payload module and for system testing at spacecraft level , thales alenia space ( turin ) responsible for the service module , and astrium ( toulouse ) responsible for the telescope , with in excess of a hundred sub - contractors .
pacs has been developed by a consortium of institutes led by mpe ( germany ) and including uvie ( austria ) ; ku leuven , csl , imec ( belgium ) ; cea , lam ( france ) ; mpia ( germany ) ; inaf - ifsi / oaa / oap / oat , lens , sissa ( italy ) ; iac ( spain ) .
this development has been supported by the funding agencies bmvit ( austria ) , esa - prodex ( belgium ) , cea / cnes ( france ) , dlr ( germany ) , asi / inaf ( italy ) , and cicyt / mcyt ( spain ) .
spire has been developed by a consortium of institutes led by cardiff university ( uk ) and including univ .
lethbridge ( canada ) ; naoc ( china ) ; cea , lam ( france ) ; ifsi , univ .
padua ( italy ) ; iac ( spain ) ; stockholm observatory ( sweden ) ; imperial college london , ral , ucl - mssl , ukatc , univ .
sussex ( uk ) ; and caltech , jpl , nhsc , univ .
colorado ( usa ) .
this development has been supported by national funding agencies : csa ( canada ) ; naoc ( china ) ; cea , cnes , cnrs ( france ) ; asi ( italy ) ; mcinn ( spain ) ; snsb ( sweden ) ; stfc ( uk ) ; and nasa ( usa ) .
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 .
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 .
the development of planck has been supported by : esa ; cnes and cnrs / insu - in2p3-inp ( france ) ; asi , cnr , and inaf ( italy ) ; nasa and doe ( usa ) ; stfc and uksa ( uk ) ; csic , micinn and ja ( spain ) ; tekes , aof and csc ( finland ) ; dlr and mpg ( germany ) ; csa ( canada ) ; dtu space ( denmark ) ; ser / sso ( switzerland ) ; rcn ( norway ) ; sfi ( ireland ) ; fct / mctes ( portugal ) ; and the development of planck has been supported by : esa ; cnes and cnrs / insu - in2p3-inp ( france ) ; asi , cnr , and inaf ( italy ) ; nasa and doe ( usa ) ; stfc and uksa ( uk ) ; csic , micinn and ja ( spain ) ; tekes , aof and csc ( finland ) ; dlr and mpg ( germany ) ; csa ( canada ) ; dtu space ( denmark ) ; ser / sso ( switzerland ) ; rcn ( norway ) ; sfi ( ireland ) ; fct / mctes ( portugal ) ; and prace ( eu ) .
this work received support from the alma - conicyt fund for the development of chilean astronomy ( project 31090013 ) and from the center of excellence in astrophysics and associated technologies ( pbf 06 ) . + aniano et al . , 2012 ,
apj , in press ( arxiv:1207.4186 ) + auld et al . , 2012 , mnras , in press ( arxiv:1209.4651 ) + baes m. , clemens m. , xilouris e. , et al . , 2010 , a&a , 518 , 53 + baldry k. , glazebrook k. and driver s. , 2008 , mnras , 388 , 945 + bell e. , mcintosh daniel h. , katz n. and weinberg m. , 2003 , apjs , 149 , 289 + bertin e. and arnouts s. , 1996 , a&ass , 117 , 393 + binggeli b. , et al . , 1985 ,
aj , 90 , 1681 + boselli a. et al .
, 2012 , a&a , 540 , 54 + boselli a. et al .
, 2010 , a&a , 518 , 61 + boselli a. and gavazzi g. , 2006 , pasp , 118 , 517 + ciesla l. , et al .
, 2012 , a&a , 543 , 161 + clemens m. , et al . , 2010 , a&a , 518 , 50 + condon j. , 1992,araa , 30 , 575 + corbelli e. et al . , 2012 , a&a , 542a , 32 + cortese l. , et al . , 2010 , a&a , 518 , 49 + dale d. et al . , 2012 , apj , 745 , 95 + davies j. , et al . , 2012 , mnras , 419 , 3505 + davies j. , et al . , 2011 , mnras , 415 , 1883 + davies j. , et al . , 2010 , a&a , 518 , 48 + davies j. , et al . , 2004 , mnras , 349 , 922 + davies j. , alton p. , bianchi s. and trewhella m. , 1998 , mnras , 300 , 1006 + de looze et al . , 2010 , a&a , 518 , 54 + devereux n. and young j. , 1990 , apj , 359 , 42 + dowell c. , et al . , 2010 ,
spie 7731 , 773136 +
draine b. and lazarian a. , 1998a , apj , 494 , l19 + draine b. and lazarian a. , 1998b , apj , 508 , 157 + draine b. , et al . , 2007 , apj , 663 , 866 + drinkwater et al . , 1999 ,
apj , 511 , 97 + drinkwater m. , gregg m. and colless m. , 2001 , apj , 548 , 139 + doyon r. and joseph r. , mnras , 1989 , 239 , 347 + dunne l. et al . , 2011 , mnras , 417 , 1510 + ebneter k. and balick b. , 1985 , aj , 90 , 183 + ferguson h. , 1989 , aj , 98 , 36 + ferguson h. , 1989a , ap&ss , 157 , 227 + freedman w. , et al . , 2001 ,
apj , 553 , 47 + gavazzi g. , et al . , 2008 , a&a , 482 , 43 + griffin m. et al . , 2010 ,
a&a , 518 , 3 + griffin m. et al . , 2009 , eas , 34 , 33 g + grossi m. et al . , 2010 , a&a , 518 , 52 + gunn j. and gott r. , 1972 , apj , 176 , 1 + horellou c. , casoli f. and dupraz , c. , 1995 , a&a , 303 , 361 + jones p. and mcadam w. , 1992 , apjs , 80 , 137 + jordan c. et al .
, 2007 , apjss , 169 , 213 + kent b. et al . , 2007 , apj , 665 , 15 + leitch e. et al . , 1997 , apj , 486 , l23 + martin a. , papastergis e. , giovanelli r. , haynes m. , springob , c. and stierwalt s. , 2010 , apj , 723 , 1359 + magrini l. et al .
, 2011 , a&a , 535 , 13 + mueller t. , nielbock m. , balog z. , klaas u. and vilenius e. , pacs photometer - point - source flux calibration pacs herschel document picc - me - tn-037 , v. 1.0 ( april 12 , 2011 ) + popescu c. and tuffs r. , 2002 , mnras , 567 , 221 + oosterloo t. and van gorkom , 2005 , a&a , 437 , 19o + ott s. , 2010 , in astronomical data analysis systems , aspc , ed .
y. mizumoto , k. morita and m ohishi , 434 , 139o + paolillo m. et al . , 2002 , apj , 565 , 883 + pappalardo et al . , 2012 ,
a&a , 545 , 75 + pilbratt g. et al . , 2010 , a&a , 518 , 1 + poglitsch a. et al . , 2010 ,
a&a , 518 , 2 + roussel h. , 2012 , arxiv:1205.2576 + sabatini s. et al . , 2003 ,
mnras , 341 , 981 + sandqvist a. , joersaeter s. and lindblad p. , 1995
, a&a , 295 , 585 + saunders w. et al . , 1990 , mnras , 242 , 318 + schindler s. , binggeli b. and bohringer , 1999 , a&a , 343 , 420 + scharf c. et al . , 2005 ,
apj , 633 , 154 + smith m. , et al . ,
2010 , a&a , 518 , 51 + soifer b. et al . , 1987 , apj , 320 , 238 + swinyard b. et al . , 2010 , a&a , 518 , 4 + taylor r. , 2010 , phd thesis , cardiff university , uk .
+ temi p. , brighenti f. and matthews w. , 2009 , apj , 707 , 890 + thomas p. , drinkwater m. and evstigneeva e. , 2008 , mnras , 389 , 102 + tuffs r. et al .
, 2002 , apjs , 139 , 37 + veron - cetty m. and veron p. , 2006 ,
a&a , 455 , 773 + waugh m.et al . , 2002 , mnras , 337 , 641 + young l. et al . , 2011 , mnras , 414 , 940 + | we present herschel space telescope observations of the nearby fornax cluster at 100 , 160 , 250 , 350 and 500@xmath0 m with a spatial resolution of 7 - 36 arc sec ( 10"@xmath11 kpc at @xmath2 17.9 mpc ) .
we define a sample of eleven bright galaxies , selected at 500@xmath0 m , that can be directly compared with our past work on the virgo cluster .
we check and compare our results with previous observations made by iras and planck , finding good agreement .
the far - infrared luminosity density is higher , by about a factor of three , in fornax compared to virgo , consistent with the higher number density of galaxies .
the 100@xmath0 m ( 42.5 - 122.5 @xmath0 m ) luminosity is two orders of magnitude larger in fornax than in the local field as measured by iras .
we calculate stellar ( @xmath3 ) and far - infrared ( @xmath4 ) luminosities for each galaxy and use this to estimate a mean optical depth of @xmath5 - the same value as we previously found for virgo cluster galaxies . for ten of the eleven galaxies ( ngc1399 excepted ) we fit a modified blackbody curve ( @xmath6 ) to our observed flux densities to derive dust masses and temperatures of @xmath7 m@xmath8 and t=14.6 - 24.2k respectively , values comparable to those found for virgo . the derived stars - to - gas(atomic ) and
gas(atomic)-to - dust ratios vary from 1.1 - 67.6 and 9.8 - 436.5 respectively , again broadly consistent with values for virgo .
fornax is a mass overdensity in stars and dust of about 120 when compared to the local field ( 30 for virgo ) .
fornax and virgo are both a factor of 6 lower over densities in gas(atomic ) than in stars and dust indicating loss of gas , but not dust and stars , in the cluster environment .
we consider in more detail two of the sample galaxies . as the brightest source in either fornax and virgo ,
ngc1365 is also detected by planck .
the planck data fit the pacs / spire sed out to 1382@xmath0 m with no evidence of other sources of emission ( spinning dust , free - free , synchrotron ) . at the opposite end of the scale ngc1399
is detected only at 500@xmath0 m with the emission probably arising from the nuclear radio source rather than inter - stellar dust .
galaxies : ism - galaxies : custers individual : fornax - galaxies : general : ism |
given three coordinate hyperplanes @xmath3 , @xmath4 , @xmath5 in @xmath0 , namely @xmath6 and smooth functions @xmath7 , @xmath8 , consider estimates of the form @xmath9 since @xmath10 by duality the above estimate is equivalent to @xmath11 by hlder s inequality , we obtain the necessary and sufficient conditions @xmath12 . in that case
we obtain the bound @xmath13 which is the classical loomis - whitney inequality in three space dimensions , see @xcite . the question which we address here
is the following : does the estimate remain true if @xmath3 , @xmath4 and @xmath5 are bounded subset of transversal , sufficiently smooth , and oriented surfaces in @xmath0 ? this question has been answered in the affirmative in ( * ? ? ?
* proposition 7 ) , along with a quantitative estimate , under the assumption of @xmath14 regularity and a local transversality condition on the surfaces . in considering this question
we are motivated by problems which arise in the analysis of bilinear @xmath15 estimates in various nonlinear dispersive equations .
precisely , one can view the estimate as a limiting case of the following bound : @xmath16 here @xmath17 are assumed to have fourier support supported in a fixed unit ball , and the norms @xmath18 , respectively @xmath19 are defined by @xmath20 respectively @xmath21 by rescaling , this implies estimates on dyadic frequency scales , in the low modulation region
. bounds of this type have already appeared at least implicitely in the study of bilinear interactions in many semilinear equations with nontrivial resonance sets , i.e. when bilinear interactions of solutions to the linear homogeneous equation have an output near the characteristic set , which in our context means that @xmath22 has a nontrivial intersection with @xmath5 .
for instance , in the context of schrdinger equations we can mention @xcite , ( * ? ? ?
* lemma 4.1 ) and @xcite ; there the three surfaces are ( pieces of ) parabolas .
other examples are the bounds for the kp - i equation considered in @xcite . a large class of bilinear and multilinear estimates
have been systematically studied in @xcite ; however , this does not include the present setup . in most applications to dispersive equations bilinear estimates
are proved in an ad - hoc manner , taking advantage of the exact form of the surfaces @xmath3 , @xmath4 and @xmath5 . in all cases mentioned above the three surfaces have nonvanishing curvature , and
one may ask which ( if any ) is the role played by the curvature in general .
it is well known that the nonvanishing curvature plays a fundamental role in the study of nonlinear dispersive equations , as it insures good decay properties for the fundamental solution of the corresponding linear equation , as well as strichartz and other estimates for solutions to the linear equation . on the other hand ,
the bound is still valid when @xmath3 , @xmath4 and @xmath5 are transversal planes .
note that the role of curvature and transversality has been clarified in @xcite in a much broader context : curvature is dispensable for the validity of estimate .
however , it still is desireable to gain a better understanding of the interplay of the necessary size , regularity , transversality and curvature assumptions on the surfaces under which sharp quantitative and scaleable estimates of the type hold true .
our main motivation are bilinear estimates with applications to the 2d zakharov system , see @xcite , in which case we need to analyze the interaction between two paraboloids and a cone .
our precise set - up is the following .
[ ass ] for @xmath23 there exists @xmath24 , @xmath25 , and @xmath26 as well as and condition [ it : reg_cond ] only to ensure the existence of a global representation of @xmath27 as a graph over a cube in a suitable frame .
this assumption is likely to be redundant , but we will not pursue these matters here as it is irrelevant for applications . ]
@xmath28 , such that @xmath27 is an open and bounded subset of @xmath29 and 1 .
[ it : reg_cond ] the oriented surface @xmath29 is given as @xmath30 for a convex @xmath31 such that @xmath32 ; 2 .
[ it : hoeldercond ] the unit normal vector field @xmath33 on @xmath29 satisfies the hlder condition @xmath34 3 .
[ it : trans ] the matrix @xmath35 satisfies the transversality condition @xmath36 for all @xmath37 .
( 0,0 ) three surfaces @xmath3 , @xmath4 , and @xmath5 in a transversal position.,title="fig:",width=415 ] ( 11,6)(0.1,0.1 ) ( 2.6,3)(0,0)[lb ] ( 7.8,1.8)(0,0)[lb ] ( 7,5.6)(0,0)[lb ] we identify @xmath38 @xmath39 being the surface measure on @xmath3 with the distribution @xmath40 for @xmath41 with compact support the convolution @xmath42 is given by @xmath43 a - priori this convolution is an integrable function which is only defined almost everywhere , therefore its restriction to @xmath5 is not well - defined . to address this issue we begin with @xmath44 and @xmath45 .
then @xmath46 and has a well - defined trace on @xmath5 .
if is proved in this case , then the trace of @xmath47 on @xmath5 can be defined by density for arbitrary @xmath48 and @xmath49 . transferring the bound from coordinate planes to
the general setting of possibly curved surfaces turns out to be quite nontrivial .
the reason is that the convolution has an additive structure with respect to addition in the ambient space @xmath0 , which is lost when restricting it to curved surfaces .
our first result is the following .
[ thm : conv ] let @xmath50 be surfaces in @xmath0 which satisfy assumption [ ass ] with parameters @xmath24 , @xmath51 and @xmath52 , and @xmath53 .
then for each @xmath54 and @xmath55 the restriction of the convolution @xmath56 to @xmath5 is a well - defined @xmath57-function which satisfies @xmath58 where the constant @xmath59 depends only on @xmath60 . as mentioned above , in the case where the surfaces are of class @xmath61 and a local transversality condition is satisfied
a variant of this theorem has been obtained earlier with a different proof . more precisely ,
estimate can be derived from ( * ? ? ?
* theorem 1 ) as in the proof of ( * ? ? ? * proposition 7 or theorem 8) . for multilinear estimates with applications to
nonlinear dispersive equations it is necessary to make explicit how @xmath59 depends on the diameter of the surfaces and on @xmath62 , @xmath63 and @xmath60 .
the subsequent corollaries [ cor : bt ] , [ cor : conv_gen ] and in particular [ cor : conv_theta ] are quantitative refinements of the above theorem which to the best of our knowledge are not available in the literature .
the proof of theorem [ thm : conv ] presented here , which merely uses @xmath2 regularity , is based on the induction on scales argument la bourgain , wolff , tao seems to be more robust compared to the proof given in @xcite in the sense that it does not require the normals to be lipschitz . on the other hand ,
the induction on scales machinery has been implemented in @xcite in a more general context , but the results of @xcite imply our results only up to a small loss in the induction on scales procedure , see also ( * ? ? ?
* remark 6.3 ) .
note that the homogeneous regularity assumption allows us to take advantage of the isotropic scaling , which is essential for the derivation of the subsequent corollaries .
the result can be viewed as a weaker form of the three dimensional multilinear restriction conjecture , see @xcite .
denoting @xmath64 we have @xcite assume that @xmath3 , @xmath4 and @xmath5 satisfy the transversality condition [ it : trans ] above .
then @xmath65 with an @xmath66 loss this is proved in @xcite , @xmath67 another generalization of the loomis - whitney inequality is given in @xcite . in the context of the above restriction conjecture ,
the results in @xcite imply that @xmath68 which would follow from by multilinear interpolation with the trivial @xmath69 bound for the product .
we conclude the section with a discussion of further versions of our main result . partitioning the three surfaces into smaller pieces and using linear changes of coordinates
it is easy to allow arbitrary values for the parameters in the hypothesis of the theorem : [ cor : bt ] let @xmath50 be surfaces in @xmath0 which satisfy assumption [ ass ] with parameters @xmath24 , @xmath25 , @xmath26 , and @xmath70 .
then for each @xmath71 and @xmath55 the restriction of the convolution @xmath56 to @xmath5 is a well - defined @xmath57-function which satisfies @xmath72 here the expression @xmath73 appears due to isotropic scaling .
while this is easy to prove , it is not so useful due to the unspecified dependence of the constant @xmath59 on @xmath73 and @xmath62 .
a better result is contained in the next corollary , which considers the case of three surfaces which can be placed into the context of theorem [ thm : conv ] via a linear transformation .
[ cor : conv_gen ] assume that @xmath3 , @xmath4 , @xmath5 satisfy the conditions of theorem [ thm : conv ] .
let @xmath74 be an invertible , linear map and @xmath75 .
then for functions @xmath76 and @xmath77 the restriction of the convolution @xmath78 to @xmath79 is a well - defined @xmath80-function which satisfies @xmath81 where @xmath82 and @xmath83 is the matrix of the unit normals to @xmath84 at @xmath85 and @xmath59 depends only on @xmath60 .
we remark that the linear transformation @xmath86 does not explicitely appear in the estimate . instead , the size @xmath87 of the constant is determined only by the transversality properties of the surfaces @xmath88 .
hence the best way to interpret the result in the corollary is to say that the bound for the surfaces @xmath88 holds whenever these surfaces are bounded , @xmath2 regular and uniformly transversal with respect to some linear frame . finally , let us state an explicit condition which guarantees that the assumptions of corollary [ cor : conv_gen ] are satisfied : [ cor : conv_theta ] let @xmath50 be surfaces in @xmath0 which satisfy assumption [ ass ] with parameters @xmath89 , @xmath25 , @xmath26 , and @xmath70 , so that @xmath90 then for each @xmath54 and @xmath55 the restriction of the convolution @xmath56 to @xmath5 is a well - defined @xmath57-function which satisfies @xmath91 where @xmath59 depends only on @xmath60 .
finally , we remark that the factor @xmath92 which appears in has also been obtained in ( * ? ? ?
* theorem 1.2 ) , but with an unspecified dependence of @xmath93 on @xmath63 and @xmath62 .
the aim of this section is to use linear transformations in order to derive corollaries [ cor : conv_gen],[cor : conv_theta ] from theorem [ thm : conv ] .
parts of these arguments will also be useful in the proof of theorem [ thm : conv ] .
we may assume that we have a single coordinate patch for each surface , i.e. there is a global parametrization @xmath94 , @xmath95 .
therefore , @xmath75 is parametrized by @xmath96 .
for @xmath23 we define the induced normals @xmath97 it is easily checked that @xmath98 and therefore @xmath99 let @xmath76 , @xmath77 and @xmath100 be given and it is enough to consider non - negative functions .
we write @xmath101 , and @xmath102 and similarly for @xmath103 , @xmath104 . with this notation and by duality , our claim is equivalent to @xmath105 with the measure @xmath106 where @xmath107 denotes the dirac delta distribution at the origin in @xmath108 .
let us also define the measure @xmath109 using @xmath110 , and we have that @xmath111 by the transversality assumption with @xmath112 on the normals to @xmath27 . therefore is equivalent to @xmath113 observe that it follows from our assumption that the corresponding estimate for @xmath50 , namely @xmath114 holds true .
this implies . partitioning each of the three surfaces into smaller sets we strenghten the relation to @xmath115 consider a fixed triplet @xmath116 . for arbitrary @xmath117 we use the hlder condition to compute @xmath118
this implies that @xmath119 does not vary much , @xmath120 hence , after possibly increasing @xmath62 , we may assume that on @xmath121 we have the stronger bound @xmath122 from the hlder condition we also obtain @xmath123 which shows that @xmath27 is contained in an infinite slab of thickness @xmath124 with respect to the @xmath125 direction . by orthogonality with respect to such slabs
it suffices to prove the desired bound in the case when the other two surfaces are contained in similar slabs , @xmath126 we will apply corollary [ cor : conv_gen ] with @xmath127 we need to show that for @xmath128 we have the conditions with @xmath51 , with @xmath52 and the size condition @xmath129 . concerning the latter we observe that @xmath130 thus by we obtain @xmath131 therefore @xmath129 .
for the transversality condition , we first estimate @xmath132 this gives a bound for @xmath86 , namely @xmath133 the unit normal @xmath134 to @xmath135 in @xmath136 is given by @xmath137 by construction for @xmath138 we have @xmath139 . by and it follows that @xmath140 which implies that @xmath141 this in turn yields the desired transversality condition @xmath142 finally , for the hlder condition we use and to compute @xmath143 which proves the desired bound for the first term in the hlder condition with @xmath51 .
the second term in is treated in the same way .
theorem [ thm : conv ] is obtained from uniform estimates for @xmath144 thickened surfaces @xmath145 given by @xmath146 our main technical result is the following .
[ prop : eps_thick ] for all @xmath3 , @xmath4 , and @xmath5 with @xmath147 as in theorem [ thm : conv ] and @xmath54 , @xmath55 and @xmath148 the estimate @xmath149 holds true with a constant @xmath59 depending only on @xmath60 .
the rest of this section is devoted to the proof of proposition [ prop : eps_thick ] . by repeating the argument from the proof of corollary [ cor : conv_theta ]
namely a finite partition of the surfaces @xmath27 , scaling and transforming the normals at one triplet of points to @xmath155 we can reduce proposition [ prop : eps_thick ] to the following setup : there are unit cubes @xmath156 , @xmath23 centered at points @xmath157 , and @xmath2 functions @xmath158 in the doubled cubes @xmath159 so that @xmath160 where the functions @xmath158 satisfy @xmath161 and have small hlder constant @xmath162 to set up the induction on scales we allow the scale of the cubes @xmath163 to vary from @xmath164 to @xmath165 .
precisely , for @xmath166 we denote by @xmath167 the best constant @xmath59 in the estimate @xmath168 considered over all surfaces @xmath3 , @xmath4 , @xmath5 as in with @xmath163 cubes of size @xmath169 and @xmath158 satisfying and . for @xmath170
we also introduce the auxiliary notation @xmath171 as the best constant in the estimate over all surfaces @xmath3 , @xmath4 , @xmath5 as in with @xmath163 cubes of size @xmath169 and @xmath158 satisfying in larger cubes @xmath172 and a weaker version of , namely @xmath173 throughout this paper ( and hence in the above definitions ) we agree to the following convention : the size of a cube is half of its side - length .
the reason for doing so is purely technical as it spares us from carrying a factor of @xmath174 in some estimates . for @xmath54 , @xmath178 and
an arbitrary test function @xmath179 the convolution @xmath180 can be expressed in the form @xmath181 where @xmath182 assume that @xmath183 .
for every @xmath184 there is @xmath185 such that @xmath186 and @xmath187 .
using we have : @xmath188 then , @xmath189 let us denote @xmath190 $ ] .
the cauchy - schwarz inequality shows that @xmath191 where @xmath192 by the change of variables @xmath193 we obtain @xmath194 because @xmath195
. finally , since @xmath196 is in an interval of size @xmath197 , using again the cauchy - schwarz inequality we obtain @xmath198 hence it follows that @xmath199 the bound @xmath200 follows by density and duality .
the previous lemma shows that @xmath167 is finite . by the argument in the proof of corollary [ cor : conv_theta ] it also follows that @xmath171 is always finite .
the next lemma allows us to bound the auxiliary variable @xmath171 in terms of @xmath167 .
the argument proceeds along the lines of the proof of corollary [ cor : conv_theta ] , carefully keeping track of the scales .
let @xmath202 be the images of @xmath203 .
we consider the linear transformation defined by the matrix @xmath204 we denote @xmath205 .
we will show that the surfaces @xmath135 satisfy the conditions in the definition of @xmath206 .
we denote @xmath207 and let @xmath208 be the projections of @xmath209 on the coordinate planes , and @xmath210 the corresponding projections of @xmath135 . setting @xmath211 for convenience , we also consider the full correspondence @xmath212 given by @xmath213 and similarly for @xmath214 . we claim that @xmath210 is contained in a cube of size @xmath222 centered at @xmath208 .
set @xmath211 for convenience , and consider the canonical map @xmath223 from @xmath224 to @xmath225 , defined by @xmath226 .
then , by and the chain rule shows that @xmath227 this implies that @xmath228 is contained inside a cube centered at @xmath218 which has size @xmath229 or next goal is to establish the bound for @xmath230 . define the function @xmath231
. then @xmath232 iff @xmath233 .
the implicit function theorem guarantees the existence of @xmath234 which satisfies @xmath235 .
in addition , setting @xmath236 , we have @xmath237 by and we obtain @xmath238 which after some elementary computations leads to @xmath239 on the other hand shows that @xmath240 given the value of @xmath241 it follows that @xmath242 hence is established for the surfaces @xmath135 .
there is a small variation in the thickness of the third surface .
a direct computation based on gives @xmath244 moreover , if @xmath245 , then @xmath246 from all the above considerations it follows that @xmath247 and the bound follows immediately since @xmath248 .
( symmetrization ) _ the numbers @xmath252 and @xmath253 are defined by ( with the additional constraints on @xmath27 ) .
that formula has the disadvantage of not revealing the symmetry of the problem with respect to the role of the three surfaces .
however , is equivalent to @xmath254 for all @xmath255 . upon replacing @xmath256 by @xmath257 and @xmath5 by @xmath258 and considering smooth , compactly supported @xmath256 , this coincides with the triple convolution of the distributions @xmath259 , @xmath260 with @xmath256 at zero , i.e. @xmath261 by density
this is an equivalent definition of @xmath262 . since is symmetric , which is irrelevant at this scale . ] in @xmath27 we prefer to use this as a definition of @xmath262 . in a similar way we symmetrize the definition of @xmath263 .
( reduction of scales ) _ from now on we assume that @xmath27 are defined as in where @xmath163 are cubes of size @xmath264 . by translation in the coordinate directions we may assume that @xmath265 , @xmath23 . from it follows for @xmath266 that @xmath267 from @xmath249 it follows that @xmath268 , therefore @xmath269 hence we are dealing with three scales ordered as follow : @xmath270 _ step 3 .
( decomposition of the surfaces ) _ inspired by we make the following construction .
we will recursively construct an increasing sequence and @xmath260 , but the final bound will not depend on this sequence . ]
@xmath271 with the properties @xmath272\label{eq : claima}\ ] ] and @xmath273 in order to do so , we set @xmath274 for a number @xmath275 such that @xmath259 and @xmath260 have support in the slab @xmath276 $ ]
. then is trivially verified for @xmath277 .
assume we already have constructed @xmath278 for some @xmath279 .
the set @xmath280 comprises @xmath281 slabs @xmath282 of thickness ( in the third coordinate direction ) @xmath283 , where @xmath281 denotes the largest integer which is smaller or equal to the ratio @xmath284 . for the function
@xmath285 it follows that @xmath286 this estimate implies that there exists @xmath287 such that @xmath288 we choose @xmath289 to be the center of @xmath287 , which satisfies because @xmath290 $ ] and because @xmath291 . for this sequence
we define @xmath292 & = \sigma_{i } \cap \ { ( x_1,x_2,x_3 ) \mid ( -1)^i x_3 \in [ s_k+r_\beta , s_{k+1}-r_\beta ] \},\\ \tilde{\sigma}_{i}[k , \mathfrak{e}_3 ] & = \sigma_{i } \cap \ { ( x_1,x_2,x_3 ) \mid ( -1)^i x_3 \in [ s_k - r_\beta , s_k+r_\beta ] \}. \end{aligned}\ ] ] with this notation it follows that @xmath293 ) } \leq 2^{-17 } r^{\frac\beta4 } \| f_i \|_{l^2(\sigma_{i}[k-1 , \mathfrak{e}_3]\cup \tilde{\sigma}_{i}[k , \mathfrak{e}_3])\cup \sigma_{i}[k , \mathfrak{e}_3])}\ , , i=1,2 .
\end{aligned}\ ] ] since has the ( more restrictive ) analog in all directions , we can perform a similar construction to define @xmath294 $ ] and @xmath295 $ ] for @xmath296 as well as @xmath297 $ ] and @xmath298 $ ] for @xmath299 with the same properties and .
moreover , we introduce @xmath300 \cap \sigma_1[k_3,\mathfrak{e}_3],\\ \tilde{\sigma}_{1}^{k_2,k_3}= { } & ( \tilde{\sigma}_1[k_2,\mathfrak{e}_2 ] \cap \sigma_1[k_3,\mathfrak{e}_3 ] ) \cup ( \sigma_1[k_2,\mathfrak{e}_2 ] \cap \tilde{\sigma}_1[k_3,\mathfrak{e}_3 ] ) \\ & \cup ( \tilde{\sigma}_1[k_2,\mathfrak{e}_2 ] \cap \tilde{\sigma}_1[k_3,\mathfrak{e}_3 ] ) , \end{aligned}\ ] ] and similarly @xmath301 , @xmath302 and @xmath303 , @xmath304 .
now , we have the decompositions @xmath305 and the same for @xmath145 .
\a ) diameter : from it follows that @xmath306 , is generated as in by @xmath307 , where @xmath59 is a cube of size @xmath93 .
in addition , since @xmath308 , it follows that at the center @xmath309 of @xmath59 we have an estimate of type , namely @xmath310 a similar characterization holds true for @xmath311 , etc .
this basically says that if in we replace each @xmath27 by @xmath312 or @xmath313 , then the constant should be adjusted to @xmath314 . indeed , by definition of @xmath317 and @xmath318 we have @xmath319 \}.
\end{aligned}\ ] ] for the left hand side of to be different from zero it is necessary that @xmath320\cap [ -r_\beta , r_\beta ] \not=0.\ ] ] due to , which leads to @xmath321 because @xmath322 is strictly increasing . in a similar manner
it follows that @xmath316 for @xmath323 .
a similar argument , using the properties of @xmath324 , provides that if one allows in one or more of the @xmath27 to be replaced by the corresponding set @xmath135 , the convolution is zero unless @xmath325 .
\c ) smallness on @xmath326 : the lack of perfect orthogonality in when @xmath326 s are involved is compensated by the following smallness of mass on those sets @xmath327 we prove for @xmath328 since the other cases are similar . from the definitions of the sets we have the straightforward estimate @xmath329)}^2 + \sum_{k_3 } \| f_1 \|_{l^2(\tilde{\sigma}_1[k_3,\mathfrak{e}_3])}^2\ ] ] then one uses and the analog of for @xmath330 $ ] to estimate each term by @xmath331 and obtains . where @xmath333 where the remainder @xmath86 contains @xmath334 sums of the same type as @xmath335 , except that one ( @xmath336 cases ) , two ( @xmath336 cases ) or all three ( @xmath165 case ) @xmath337 are replaced by @xmath326 .
we decompose as @xmath338 where the subscripts indicate the positions of the @xmath326 factors . from the conclusions in step 4
a ) above , we obtain @xmath340 in passing to the last line we have used the cauchy - schwartz inequality with respect to all three @xmath341 s and the fact that the sets @xmath317 are disjoint with respect to the pair @xmath342 ( and the same for the sets @xmath301 , @xmath343 ) .
for each term in the remainder the same argument applies up to the orthogonality issue . by the almost orthogonality the first term in the remainder
is given by @xmath344 where @xmath345 indicates summation with respect to @xmath346 satisfying @xmath347 for @xmath23 .
we obtain @xmath348 where we have used in passing to the last line . if one considers the remaining terms in the same estimate holds true , which amounts to @xmath349 this estimate for the remainder term @xmath86 together with the estimate for the main term @xmath335 and leads to .
we recursiveley define @xmath353 as @xmath354 a straightforward computation gives @xmath355 since the right - hand side is an increasing function in @xmath353 , one can choose @xmath356 such that additionally @xmath357 is satisfied . with this choice the sequence @xmath353 is strictly decreasing and @xmath358 .
the result in provides @xmath359 however , in order to apply the above inequality we need to verify the required bounds . if @xmath360 then the above formula can be used as long as @xmath361 . for @xmath362 one
needs to verify @xmath363 for which it is enough to take @xmath364 .
for @xmath279 we have that @xmath365 is an increasing sequence , therefore we can find @xmath366 to be the highest value of @xmath367 with the property @xmath368 .
now we can use for all @xmath369 and by iterating it , we obtain @xmath370 using we estimate @xmath371 which is less than @xmath372 by making @xmath373 small enough , which shows that @xmath374 and therefore @xmath375 for all @xmath376 .
since @xmath377 , it follows that @xmath378 .
now we can apply the result in lemma [ lem : start ] and obtain @xmath379 in a similar manner as above we obtain @xmath380 at the expense of adjusting @xmath373 again . the last estimate together with and proves that @xmath381 for all @xmath382 . from this
the claim in proposition [ prop : eps_thick ] follows by partitioning each @xmath27 into a finite number of pieces of diameter less than @xmath383 and , in addition , by partitioning @xmath145 into a finite number of pieces @xmath384 where @xmath79 are translates of @xmath5 in the @xmath196-direction . | given three transversal and sufficiently regular hypersurfaces in @xmath0 it follows from work of bennett carbery
wright that the convolution of two @xmath1 functions supported of the first and second hypersurface , respectively , can be restricted to an @xmath1 function on the third hypersurface , which can be considered as a nonlinear version of the loomis
whitney inequality .
we generalize this result to a class of @xmath2 hypersurfaces in @xmath0 , under scaleable assumptions .
the resulting uniform @xmath1 estimate has applications to nonlinear dispersive equations . |
the microscopic description of the surface of liquids touches various fields of natural science .
this problem is not simple , and even the calculation of the surface tension coefficient has many difficulties.@xcite at low temperature the quantum nature of the surface excitations becomes important . at very low temperature only few types of surface excitations with the energy less or of the order of temperature
are relevant for the problem .
this fact greatly simplifies the description of liquid surface in the low - temperature limit .
an accurate calculation of the absolute value of the surface tension coefficient @xmath0 remains a challenge , but the calculation of its temperature dependence @xmath1 is much simpler . among all liquids ,
the low - temperature limit is reached only in helium , and the surface of liquid helium has been studied experimentally in great detail.@xcite therefore , we apply our analysis mainly to the liquid helium .
the experimental values of the surface tension of liquid @xmath2he and @xmath3he at zero temperature aresurtensexp4,surtensexp3 @xmath4dyn / cm and @xmath5dyn / cm . taking these values as a reference point
, one may calculate the deviation @xmath6 @xmath7 as the sum of the contributions from all types of surface excitations to the surface free energy per unit square . at low temperature
the concentration of these excitations is low , and the interaction between the excitations can be neglected .
then the surface excitations form an ideal two - dimensional gas of particles with the dispersion determined by the nature of these excitations .
the free energy of this gas is well known@xcite [ see eq . ( [ omsigma ] ) below ] . at sufficiently low temperatures ,
the only considered type of surface excitations are the quanta of surface waves , called ripplons .
the ripplons lead to the temperature dependence of the surface tension of a liquid given by the atkins formula @xcite:@xmath8where the coefficient @xmath9 mdyn/@xmath10k@xmath11 for @xmath2he .
however , this estimate of the temperature dependence of the surface tension is much weaker than the measured one.@xcite moreover , eq .
( atkins ) applies only for @xmath2he in the superfluid state , because above the @xmath12-point , @xmath13k , the short - wavelength ripplons are damped by the liquid viscosity .
the damping of the high - energy short - wavelength ripplons must lead to the strong modification of the dependence ( [ atkins ] ) above the @xmath12-point . on contrary
, experiment@xcite gives only very weak change in the temperature dependence of the surface tension at the @xmath12-point , suggesting that the ripplon contribution is not the main one . in liquid
@xmath3he the viscosity is much higher than in @xmath2he , and only the very long - wavelength ripplons with energy @xmath14 survive at @xmath15 .
these long - wavelength ripplons give negligible contribution to the free energy because of the small number of quantum states , which is proportional to the phase volume .
hence , the theory predicts a very weak dependence @xmath1 for @xmath3he in the whole temperature interval , which strongly contradicts the experimental observations@xcite .
this discrepancy between the theory and experiment on the liquid helium surface tension remained a puzzle for several decades , until the new type of surface excitations has been proposed.@xcite this new type of excitations , called the surface level atoms ( sla ) or _ surfons _ , allowed to explain the temperature dependence of the surface tension and to reach the very high agreement between theory and experiment.@xcite the surfons resemble the states of @xmath3he atoms on the surface of liquid @xmath2he in the @xmath3he-@xmath2he mixtures @xcite and the states of @xmath3he or @xmath2he atoms on the surface of liquid hydrogen.@xcite the surface states in the @xmath3he-@xmath2he mixtures , called the andreev states , were also introduced to explain the temperature dependence of the surface tension of these mixtures .
the main difference between the andreev states and the surfons is that the latter exist even in the pure he isotopes .
a thorough microscopic description of this new type of excitations is a rather complicated many - particle problem . in ref .
@xcite the surfons were considered phenomenologically as the quantum states of helium atoms localized at the liquid helium surface .
the surfons may propagate in the surface plane and have the quadratic dispersion@xmath16where @xmath17 is the 2d momentum of surfons along the surface .
both the surfon energy @xmath18 and their effective mass @xmath19 depend on the he isotope @xmath3he or @xmath2he .
the surfon energies @xmath18 are intermediate between the energy of a he atom in vacuum @xmath20 and the chemical potential @xmath21 of this atom inside the liquid .
if one takes the energies of he atoms in vacuum as the reference point , @xmath22 , the chemical potentials of liquid @xmath2he and @xmath3he at @xmath23 are @xmath24the creation of a surfon is a thermal activation process with the activation energy @xmath25therefore , at low enough temperature @xmath26 the concentration of surfons is exponentially small . fitting the experimental data on the temperature dependence of the surface tension , below we obtain the values of this activation energy for @xmath2he and @xmath3he : @xmath27the corresponding energies of surfons are @xmath28the effective masses , obtained from this fitting , are @xmath29where @xmath30 and @xmath31 are the masses of free @xmath2he and @xmath3he atoms .
the role of surfons may be crucial for various properties of electrons on the liquid helium surface ( see refs .
@xcite for reviews of this area ) .
electron scattering on surfons reduces the mobility of the electrons.@xcite this scattering also affects the electron transitions between the bound states on the surface , leading to the shift and the broadening of the transition line.@xcite this is particularly important for the physical realization of quantum bits and quantum computing with the use of electrons on the liquid helium surface.@xcite in the present paper we microscopically substantiate the existence of surfons and study their properties .
we estimate the energy ( [ slaenergy ] ) ant the effective mass ( [ m ] ) of surfons from the microscopic considerations . we show that taking into account the temperature dependence of the chemical potential of liquid helium allows to reach the agreement between theory and experiment on the dependence @xmath32 up to the accuracy of the existing experiments .
we propose the theoretical curve of this dependence in the entire temperature interval from zero to the boiling temperature @xmath33 of liquid helium .
this theoretical prediction explains the existing puzzles in the temperature dependence of the surface tension of liquid helium and other liquids , and stimulates further experiments .
the goal of this section is to prove the existence of the surfons from the quantum - mechanical calculation and to give a rough estimate of its activation energy and effective mass . in the next section
we find these values phenomenologically with higher accuracy from the comparison with experiment on the temperature dependence of the surface tension @xmath34 . in this section
we use the simplified interaction between he atoms , which includes only the hartree term of the van der waals forces and neglects the exchange interaction .
then we solve the one - particle schrdinger equation for the surface atom in the potential , formed by other atoms , taking into account possible formation of a dimple ( or polaron ) under the atom on the surface .
the interaction potential between two helium atoms can be described by the lennard - jones ( lj ) potential@xcite @xmath35 , \label{lj}\]]where the generally accepted coefficients @xmath36are the same for both he isotopes and obtained by fitting the he atom scattering experiments .
this potential includes the van - der - waals attraction between atoms at long distance and the hard - core repulsion at short distance .
more complicated he - he interatomic potentials have been proposed by various authors,@xcite but for our estimates the accuracy of the potential ( [ lj ] ) is safficient .
the potential energy of an atom above the liquid can be calculated by the summation of the interatomic potential over all atoms in the liquid .
this approximation neglects the many - particle effects and the back influence of an atom above the surface on the bulk liquid , studied later .
the integration of the lennard - jones potential ( lj ) over the uniform liquid in the half - space @xmath37 gives the following potential energy of an atom above the surface of liquid:@xmath38 .
\label{vz}\]]here @xmath39 is the distance of the atom from the surface , @xmath40 is the number atom density in the bulk liquid .
this bulk density in liquid @xmath2he at @xmath41 is @xmath42 , and in @xmath3he @xmath43 .
the potential ( [ vz ] ) is attractive at long distance @xmath44 and repulsive at @xmath45 .
the substitution of the numbers to eq .
( [ vz ] ) gives the following potentials of he atoms above @xmath46he and @xmath3he surfaces:@xmath47 , \label{vhe4 } \\
v^{he3}\left ( z\right ) & = & \left ( \frac{3640}{z^{9}}-\frac{97.9}{z^{3}}% \right ) \left [ k\right ] .
\label{vhe3}\end{aligned}\]]the numerical solution of the one - dimensional schr " odinger equation for a he atom in these potentials * proves the existence * of discrete energy levels for both he isotopes , and gives the following estimates for the energy of this bound state in the zeroth approximation : @xmath48these calculated values are higher than the values ( [ slaenergy ] ) obtained from the surface tension data fit , which is not surprising .
first , the above model takes the liquid surface to be rigid and flat , while in fact this surface is very soft and subjected to deformation , which reduces the bound state energy ( [ slaenergy0 ] ) .
second , the distribution of he atoms in the liquid is not uniform and is affected by an atom on the surface level .
one expects the increase of liquid density around the surfons.third , the identity of the atom on the surface level with the atoms in the bulk liquid leads to the exchange energy correction and other many - body effects.@xcite the similar problem appears in the calculation of the energy of andreev levels,@xcite i.e. of the bound states of @xmath3he atoms on the surface of liquid @xmath2he .
the solution of 1d schr " odinger equation with the potential ( [ vz ] ) gives the energy of bound state @xmath49 while the value of andreev energy levels , as obtained from the surface tension experiments , is about @xmath50 .
various approaches have been developed to calculate the andreev energy levels more accurately,@xcite but the substantial disagreement between theoretical and experimental results still exists . in this section
we study the back influence of the surfons on the bulk liquid . in the first approximation
, one must consider the formation of a dimple under the surfon , similar to the dimple under an electronshikin , edelman , monarkha or negative ion@xcite on the liquid helium surface . in the next approximation one also needs to account for the increase of liquid density in the vicinity of surfon .
this adjustment of the liquid to the appearance of the surfon is similar to the polarons around electrons in solids.@xcite the surface deformation is a polaron of ripplons , while the increase of the liquid density under the surfon is a polaron of bulk phonons .
this composite polaron propagates with the surfon along the liquid surface , reducing the surfon activation energy and increasing its effective mass .
since the relative variation of the liquid density around the surfon is small , the correction to the surfon activation energy from the density polaron is also small . on contrary , the surface deformation around the surfon may considerably change the van der waals potential at the surface , and this effect must be taken into account . for the estimate of the correction to the surfon energy and its effective mass due to the dimple on the liquid surface ,
we apply the methods similar to those in the theory of electrons@xcite and negative ions@xcite on the liquid helium surface , replacing the electrostatic attraction by the van - der waals forces .
the formation of a dimple on the surface under the surfon gains the van - der waals energy @xmath51 but costs the surface tension energy @xmath52 and the quantum kinetic energy @xmath53 of the surfon due to the additional confinement of its wave function near the dimple .
all these contributions depend on the shape of the dimple and determine this shape .
their self - consistent calculation requires the solution of the axially - symmetric 3d schr " odinger equation for he atom above the dimple .
we do not perform the complicated calculation of the dimple profile , because the model of a classical static surface deformation itself is rather rough . instead , we estimate the energy gain due to the formation of the dimple taking its shape @xmath54 to be a spherical cap of radius @xmath55 and depth @xmath56 : @xmath57the loss in the surface tension energy @xmath52 of the shallow dimple of depth @xmath58 is approximately proportional to @xmath59 and depends weakly on the shape of the dimple .
thus , for the spherical dimple ( [ xi1 ] ) @xmath60while for the much smoother gaussian shape @xmath61 with the same depth and curvature in the center , the surface energy loss reduces by one half : @xmath62 .
the van der waals attraction is a very short - range one .
therefore , the optimal dimple does not extend far from the surfon .
the optimal dimple radius is determined from the competition between the van der waals attraction @xmath51 and the kinetic energy @xmath53 .
this competition also happens in the bulk liquid , and leads to the mean inter - particle distance @xmath63 for @xmath2he and @xmath64 for @xmath3he .
below , we take the dimple curvature radius @xmath65 in our estimate of the sla energy correction . the depth of the dimple is determined from the competition between the surface tension energy @xmath66 and the van der waals energy @xmath51 . the van der waals energy gain from the dimple ( [ xi1 ] ) can be estimated by the integration of the lj potential ( [ lj ] ) over the region @xmath67 @xmath68the kinetic energy @xmath69 already implicitly enters eq .
( [ ew ] ) through the surfon wave function @xmath70 and the minimal dimple radius @xmath71 .
its additional contribution reduces the energy gain ( [ ew ] ) by the factor @xmath72 .
however , we neglect this correction because we did not account for the similar factor in the surface tension energy loss .
we also reduce the van der waals energy gain by similar factor by taking the wave function in the integral ( [ ew ] ) in the form of @xmath73-function : @xmath74 , where @xmath75 is the position of the center of surfon .
thus , we have @xmath76 where @xmath77 and @xmath78 .
integration gives @xmath79 .\]]for @xmath2he this gives @xmath80 and for @xmath3he this gives @xmath81 , where the dimple depth @xmath56 is in @xmath82 and @xmath83 is in @xmath84 .
minimization of @xmath85gives for @xmath2he the optimal dimple depth and energy gain@xmath86and for @xmath3he@xmath87the surfon effective mass increases due to the dimple .
this increase can be roughly estimated assuming that during the surfon motion on the interatomic distance @xmath71 at least all adjacent atoms must also move by the distance of the order of the dimple depth .
then , the surfon effective mass @xmath88the number of adjacent atoms @xmath89 , and we get @xmath90 the above rough calculation is aimed to show that the surface deformation under the surfon is essential both for the surfon activation energy . for more accurate study of the surfon parameters , in the next section we compare with experiment the thermodynamical properties of liquid surface , calculated using the above model of surfons .
the contribution @xmath91 of the 2d surfon gas to the temperature dependence of the surface tension is@xcite @xmath92 \frac{d^{2}k}{(2\pi \hbar ) ^{2}}. \label{omsigma}\]]this is just a contribution of the 2d gas of noninteracting particles with dispersion @xmath93 to the free energy of unit surface area.ll5 here @xmath94 is the temperature - dependent chemical potential of liquid helium , @xmath93 is the dispersion of surfons , @xmath95 is the spin degeneracy ( @xmath96 for @xmath2he and @xmath97 for @xmath3he ) , and the sign `` @xmath98 '' in ( [ omsigma ] ) is `` @xmath99 '' for bosons and `` @xmath100 '' for fermions .
the surfons above @xmath2he are the bosons with dispersion ( [ disp ] ) . after introducing the new variable
@xmath101 $ ] , their contribution to the surface tension becomes @xmath102 } { z}dz , \label{omboson}\]]where @xmath103above @xmath3he the surfons are fermions , and from ( [ omsigma ] ) we obtain after the integration by parts@xmath104 } { z}dz , \label{omfermion}\]]where @xmath105from ( [ omboson]),([omfermion ] ) we see that the contribution of surfons to @xmath106 depends exponentially on temperature at low @xmath107 . to calculate the integrals in ( [ omboson]),([omfermion ] )
one need to know the temperature dependence of the chemical potential @xmath108 , which enters eqs .
( [ omboson ] ) and ( [ omfermion ] ) via the surfon energy gap @xmath109 . in ref .
@xcite this dependence has been considered only for @xmath3he and only qualitatively using the approximate formula ( 20 ) of ref .
for quantitative analysis in the wider temperature range we take the dependence @xmath110 from the experiment .
we take the exact formula for the chemical potential @xmath111here the pressure @xmath112 is equal to the saturated vapor pressure , and@xmath113the temperature - dependent particle density @xmath114 in liquid he is determined from the mass density .
the free energy per unit volume is @xmath115where the heat capacity @xmath116 and the entropy @xmath117 at constant volume are also taken from experiment .
for @xmath2he the data on the temperature dependence of the quantities @xmath118 entering eqs .
( [ mu]),([fv ] ) are taken from ref .
for @xmath3he the data on @xmath119 and @xmath120 are taken from ref .
mutexp , and the data on saturated pressure are taken from ref .
the data on the temperature dependence of the chemical potential and of the surfon activation energy are plotted in figs .
[ fig1 ] and [ figmu3 ] .
note that for both helium isotopes the experimentally measured temperature dependence of the chemical potential is in a high agreement with the predictions of the theory of quantum nondegenerate liquids , developed in refs .
@xcite . for liquid @xmath2he divided by temperature , as it enters eq .
( [ omboson ] ) .
this dependence is extracted from the experimental temperature dependence of chemical potential @xmath121 of liquid @xmath2he , shown in the insert figure and obtained from the data in ref .
the graph shows that @xmath122 is almost independent of temperature in the wide range @xmath123 .
such a dependence of @xmath124 , being substituted in to eq .
( [ omboson ] ) , explains the quadratic temperature dependence of the surface tension at @xmath125 . ]
he ( solid line ) compared with the experimental data from ref .
@xcite ( black triangulares ) . at @xmath126
the agreement is so high that in the linear scale the deviation can not be detected by eye . in this temperature region
the ripplons are damped by viscosity , and their contribution to the surface tension @xmath127 is negligible . at low temperature ,
the concentration of surfons is exponentially small due to the activation energy @xmath128 , and the main contribution to @xmath127 comes from the gapless ripplons .
the dotted line represents eq .
( [ atkins ] ) , which gives the ripplon contribution without viscosity and is valid only deep inside the superfluid phase .
this line agrees with the experimental data only at @xmath129.,scaledwidth=49.0% ] he divided by @xmath130 .
this graph shows the comparison between our theory and experiment in the coordinates where the difference can be visually observed .
the low - temperature deviation at @xmath131 is due to the ripplon contribution , while the deviation at @xmath132 is due to the proximity to the boiling point .
the ripplon contribution at @xmath131 , extracted from this graph , is shown in fig .
[ figripplons],scaledwidth=49.0% ] to the surface tension of @xmath2he , given by eq .
( [ omboson ] ) , and the experimental values @xmath133 .
the maximum at @xmath134 is due to the ripplon contribution .
the second maximum at @xmath135 corresponds to phase transition at the @xmath136-point of @xmath2he.,scaledwidth=49.0% ] in the superfluid @xmath2he at @xmath137 the chemical potential @xmath138 depends weakly on temperature . at @xmath139
this dependence is quadratic ( see insert of fig .
[ fig1 ] ) in accordance with the theoretical prediction@xcite @xmath140where @xmath141 and @xmath142 .
the function @xmath143 , which determines the temperature dependence of the surface tension @xmath144 in ( [ omboson ] ) and is given by eq .
( [ deltat4 ] ) , depends weakly on temperature in the interval @xmath145 ( see fig . [ fig1 ] ) .
the minimum of the function @xmath122 occurs at @xmath146 , but in the entire range @xmath147 one can take @xmath148 with the accuracy @xmath149 .
then the integrand in eq .
( [ omboson ] ) does not depend on temperature , which results in the quadratic temperature dependence of the surface tension of liquid @xmath2he with the same accuracy @xmath150 ( see figs .
[ fig2],[fig3 ] ) .
this explains the long - standing puzzle of the experimentally observed quadratic temperature dependence of liquid @xmath2he at @xmath139 .
the values @xmath151 and @xmath152 give the best fit to the experimental points .
the ripplon contribution to the temperature dependence of the surface tension @xmath144 is small and has maximum at @xmath134 ( see fig . [ figripplons ] ) , where it makes only @xmath153 of @xmath154 ( see figs .
[ fig2],[fig3 ] ) . in ref .
surstates it was shown that at @xmath139 the thermal ripplons with energy @xmath155 , which give the main contribution to @xmath156 , are strongly damped by viscosity . in @xmath3he the short - wave - length thermal ripplons are strongly damped in the whole temperature range .
therefore , the ripplon contribution to @xmath157 is small .
the long - wave - length ripplons with energy @xmath158 are not damped by viscosity , but their contribution to @xmath159 is small because of the small phase volume .
the temperature dependence of the chemical potential @xmath160 in @xmath3he , obtained using eqs .
( [ mu]),([fv ] ) from the experimental data in ref .
@xcite , is shown in fig .
[ figmu3 ] . at @xmath161 , where @xmath3he is nondegenerate , this dependence with very high accuracy coincides with the theoretical prediction@xcite @xmath162where @xmath163 at @xmath164 , when @xmath3he is a degenerate fermi liquid , @xmath165where @xmath166
. note , that the quadratic temperature dependence of the chemical potential [ see eqs .
( [ mu4]),([mu3 t ] ) ] is the characteristic feature of all quantum nondegenerate liquids.@xcite such temperature dependence @xmath110 also takes place in liquid hydrogen.@xcite an alternative theory of this dependence is given in ref.@xcite .
similarly to the case of @xmath2he , in @xmath3he@xmath167 is almost independent of temperature with accuracy @xmath168 in the wide temperature range @xmath169 ( see insert in fig .
[ figmu3 ] ) . according to eq .
( [ omfermion ] ) , this occasional feature of the function @xmath170 results in the nearly quadratic dependence @xmath171 in the same temperature interval ( see fig .
figsig3 ) . from comparison with the experiment on @xmath172 in ref .
@xcite we obtain the values @xmath173 and @xmath174 .
one sees , that the account for the temperature dependence of the chemical potential substantially improves the agreement between the proposed theory and the experimental data on @xmath175 for both helium isotopes and extends the temperature interval of this agreement to the whole interval of the liquid phase ( see figs .
[ fig2 ] and [ figsig3 ] ) . according to the above calculations and to the experimental data , in a wide temperature range the deviation of the surface tension coefficient @xmath176 and @xmath177 .
this shows that the particle statistics of surfons is not very important for the surface phenomena on the liquid helium .
the extracted parameters of the surfon activation energy and their effective mass are in a reasonable agreement with the theoretical predictions ( [ de4]),([de3]),([m*n ] ) of our rough model and can be used for more elaborated theoretical study of the structure of this new type of surface excitations .
, obtained from the data in ref .
@xcite and plotted as function of @xmath130 to show that eq .
( [ mu3 t ] ) is fulfilled with high accuracy .
the insert figure shows the temperature dependence of the surfon gap @xmath178 divided by temperature .
one sees that @xmath179 is almost independent of @xmath180 in the wide temperature interval @xmath181 .
this dependence , substituted to eq .
( omfermion ) , explain the quadratic temperature dependence of the surface tension @xmath182 at @xmath183.,scaledwidth=49.0% ] he , calculated using eqs .
( [ omfermion]),([deltat3 ] ) ( solid line ) compared to the experimental data from ref .
surtensexp3 ( black squares ) .
the agreement is so high that in the linear scale ( insert figure ) the deviation can not be detected by eye .
therefore , on the main figure we plot @xmath184 .
the agreement between theory and experiment considerably improves after taking the experimental dependence of @xmath185 into account.,scaledwidth=49.0% ]
in sec . ii we proposed a simple theoretical model of the new type of surface excitations ( called surfons ) , and perform the quantum - mechanical calculation of the surfon activation energy @xmath18 and their effective mass @xmath19 basing on this model .
this calculation substantiates the existence of surfons , because it shows that he atoms have at least one discrete energy level at the liquid surface .
he atoms are attracted to the surface by the van der waals force and become localized in the direction , perpendicular to the surface , propagating only in the surface plane . as the result
, they form a 2d gas on the liquid surface .
the temperature - dependent concentration of the surfons is determined by their activation energy @xmath186 [ see eq .
( [ delta ] ) ] . if @xmath187 the liquid is unstable . if @xmath188 , the concentration of surfons in negligibly small compared to the concentration of helium vapor at any temperature . for both he
isotopes , the calculated values @xmath18 lie in the interval @xmath189 . the formation of a dimple under the surfon reduces the surfon activation energy by the value @xmath190 [ see eqs . ( [ de4]),([de3 ] ) ] .
the dimple also increases several times the effective mass of the surfons in their in - plane motion [ see eqs .
( [ m*]),([m*n ] ) ] .
this increase is not surprising , because during the in - plane motion the surfon also drags the dimple , which includes the motion of several atoms in the liquid .
the experimental observation and investigation of surfons is possible because the surfons make the main contribution to the temperature dependence of the surface tension . in sec .
iii we calculate this contribution , taking into account the temperature dependence of the chemical potential and liquid density , which we derive from the experimental data .
this considerably improves the agreement between our theory and experiment of the temperature dependence of the surface tension as compared to the previous lettersurstates , making the deviations to be as small as @xmath191 in the whole temperature range [ for @xmath2he this agreement is achieved after the inclusion of ripplon contribution , given by eq .
( [ atkins ] ) at temperature below the @xmath12-point ] .
the two extracted fitting parameters , the surfon activation energy @xmath109 and the effective mass @xmath192 , are in a reasonable agreement with our calculations in sec .
their values are given by eqs .
( [ d])-([m ] ) . at
very low temperature the surfon contribution to the temperature dependence of the surface tension is small , because the concentration of surfons is exponentially small being determined by the activation energy @xmath109 . at high temperature the surfon contribution @xmath193 for both helium isotopes . for even higher agreement between theory and experiment on the temperature dependence of the surface tension we also considered the interaction between surfons .
at low temperature this interaction is not important , because the concentration of surfons is low . at higher temperature the interaction between surfons only renormalizes the surfon effective mass .
the details of these results will be published elsewhere .
a.g . thanks prof . a.f .
krutov for useful discussions .
the work was supported by the foundation `` dynasty '' and by the visitor program of max planck institute for the physics of complex systems .
m. iino , m. suzuki , a.j .
ikushima and y. okuda , j. low temp . phys .
59 , 291 ( 1985 ) ; m. suzuki , y. okuda , a.j .
ikushima and m.iino , europhys .
lett . 5 , 333 ( 1988 ) ; k. matsumoto , y. okuda , m. suzuki and s. misawa , j. low temp .
125 , 59 ( 2001 ) .
r. hellmann , e. bich and e. vogel , molecular physics , * 105 * , 3013 ( 2007 ) ; r. a. aziza and m. j. slaman , j. chem .
* 94 * ( 12 ) , 8047 ( 1991 ) .
nitz , d.sieglaff , m. lagus et al .
a * 47 * , 3861 ( 1993 ) .
this exchange energy correction is expected to be small and less important for the value of the surface energy level than the local adjustment of the atoms in the liquid to the surfon .
the small value of the exchange energy comes from the very small overlap of the wave functions of he atoms due to their hard - core repulsion . | we investigate the new type of excitations on the surface of liquid helium .
these excitations , called surfons , appear because helium atoms have discrete energy level at the liquid surface , being attracted to the surface by the van der waals force and repulsed at a hard - core interatomic distance .
the concentration of the surfons increases with temperature .
the surfons propagate along the surface and form a two - dimensional gas . basing on the simple model of the surfon microscopic structure
, we estimate the surfon activation energy and effective mass for both helium isotopes .
we also calculate the contribution of the surfons to the temperature dependence of the surface tension .
this contribution explains the great and long - standing discrepancy between theory and experiment on this temperature dependence in both helium isotopes .
the achieved agreement between our theory and experiment is extremely high .
the comparison with experiment allows to extract the surfon activation energy and effective mass .
the values of these surfon microscopic parameters are in a reasonable agreement with the calculated from the proposed simple model of surfon structure . |
statistics of a set of one - dimensional random walks _ conditioned never to collide _ in a given time interval , say @xmath2 , has its own importance in statistical physics , since , if we set @xmath4 , it realizes the one - dimensional fermi statistics @xcite , and with @xmath5 it is used to analyze the models for wetting and melting phenomena @xcite .
we will refer to such non - colliding random walks and the continuum counterpart , non - colliding brownian motions , simply as _ vicious walks _ following the terminology used by m. fisher @xcite . for
the pioneering work on vicious walker models , see @xcite .
a generic setting of vicious walk problems is discussed in @xcite .
recently the interest on the vicious walks in mathematical physics is renewed and growing very rapidly , for close relationships of the vicious walk problem with the study of ensembles of young tableaux and the symmetric functions @xcite , the theories of orthogonal polynomials and random matrices @xcite , and some topics of representation theory and probability theory @xcite have been clarified . in an earlier paper @xcite ,
the continuum limit of non - colliding random walks on a lattice was taken by letting the temporal and spatial units @xmath6 go to zero with the relation @xmath7 and a system of non - colliding brownian motions was derived . since each random walk tends to be a brownian motion in this _
diffusion scaling limit _ , such a construction of non - colliding brownian motion is plausible , and indeed mathematical rigor can be established as a functional central limit theorem of vicious walks @xcite .
the important fact is that the repulsive interaction among the obtained brownian particles is no longer contact interaction as in the original vicious random walks on a lattice but is long ranged .
the origin of this long - ranged interaction is the restriction of allowed configurations by the non - colliding condition , so that we can say it is an example of entropy - origin effective force .
moreover , the following setting was considered in our vicious walks ; ( i ) even after taking the continuum limit , still we let the time interval , in which the non - colliding condition is imposed , be finite @xmath5 , and ( ii ) all the brownian particles are assumed to start from an origin and the time interval of non - colliding is @xmath0 $ ] . in this
setting the process is temporally inhomogeneous . at the very early stage @xmath8 the repulsive interaction
should be strong , since the brownian motions will be restricted so that they will not collide for a long time period up to time @xmath2 in the future . as the time @xmath1 going on , the strength of repulsion is decreasing as is the remaining time until @xmath2 , and attains its minimum at the final time @xmath9 ,
at which there is no more restriction of motion in the future @xmath10 .
it was dyson s idea that such systems of brownian motions with long - ranged repulsion could have the equilibrium states , which can be described by the distribution functions of eigenvalues of random matrices in the ensembles appropriately specified by symmetry of the system @xcite .
the previous paper @xcite showed that the spatial distribution of vicious walkers at @xmath8 realizes the eigenvalue - statistics of the gaussian unitary ensemble ( gue ) , one at @xmath9 does that of the gaussian orthogonal ensemble ( goe ) , and the gue to goe transition is observed in the intermediate time , which is equivalent with the transition studied in the two - matrix model by pandey and mehta @xcite . in the present paper
, we will characterize this transition in distribution by calculating the moments of the positions of @xmath11 particles as functions of time @xmath1 , @xmath12 where @xmath13 denotes the position of @xmath14-th vicious walker and @xmath15 the average at time @xmath16 $ ] .
in addition to such an interest of statistical physics , we can put emphasis that our present study possesses another importance as an interesting application of the graphical expansion theory . in high energy physics , graphical expansions for the matrix models of su(@xmath11 )
gauge theory were studied using _ ribbon graphs _ representing propagators and it was shown that the dominant graphs for large @xmath11 are the planar ones and the leading corrections come from the graphs embedded in a torus , which are depressed by a factor @xmath17 with respect to the planar graphs @xcite . on the other hand
, it was clarified that in the graphical expansion of so(@xmath11 ) gauge theory the leading corrections only are depressed by a factor @xmath18 for large @xmath11 with respect to the dominant planar graphs , in which propagators can be represented by _ twisted ribbons _ @xcite . in the gauge theory ,
the presence of non - gaussian interaction terms of cubic or higher power is crucial and it leads to _ triangulations of random surfaces_. although the gaussian matrix models associated with the present vicious walker model are not related with the triangulation problem of random surfaces and only give purely enumerative problems of surfaces , the proper structures of large @xmath11 expansions in su(@xmath11 ) and so(@xmath11 ) gauge theories are also found in the gue and goe models , respectively .
we can show that , if @xmath4 in our vicious walker model , the moments of the walker positions ( [ eqn : moment0 ] ) can be calculated by the graphical expansion method of gue using the ribbon graphs and the results are given in the form of power series in the inverse of the _ square _ of matrix size @xmath11 , @xmath19 here the coefficients @xmath20 are the numbers of _ orientable surfaces of genus _
@xmath21 made from @xmath22-gon by some specified procedure @xcite ( see also @xcite and sec.5.5 in @xcite ) . on the other hand , for the goe we have to take into account the _ non - orientable surfaces _ as well as the orientable ones and the expansion is in the form of power series in @xmath18 .
in other words , in order to generate necessary surfaces , we need to use twisted ribbons , whose type of graphs is now called of _ mbius graphs
for the moments ( [ eqn : moment0 ] ) of our vicious walkers , we will perform the mbius graph expansion in the present paper .
now the weights of the graphs are depending on the time @xmath1 ; in the limit @xmath3 all the weights on the twisted ribbons are zero , but they are growing as time @xmath1 going on , and at @xmath9 twisted ribbons are equally weighted as untwisted ones .
this gives another characterization of the temporally inhomogeneity of the process and the gue - to - goe transition .
quite recently nagao , tanemura and one of the present authors applied the method of skew orthogonal polynomials and quaternion determinants developed for the multimatrix models @xcite to the vicious walk problem and derived the quaternion determinantal expressions for dynamical correlation functions @xcite . using this result
, we will present an expression of the coefficients in our expansion of moments using the stirling number of the first kind .
the paper is organized as follows . in sec.ii
we briefly review the previous results reported in @xcite and give the precise definition of the moments which we will study .
graphical representations are demonstrated in sec.iii .
application of the result of @xcite is given in sec.iv to give the expression for the coefficients of expansion and some concluding remarks are given in sec.v .
appendices a and b are prepared to derive the expression of the density function used in sec.iv and the @xmath18-expansions of the one - point green function discussed in sec.v , respectively .
we study a continuum model of vicious walks , the non - colliding brownian motions in the time interval @xmath23 $ ] , constructed in @xcite as the diffusion scaling limit of vicious random walks on a lattice .
first we briefly review our previous results .
the configuration space of the present @xmath11 vicious walkers is @xmath24 , where @xmath25 is a set of all real numbers .
the probability density of vicious walkers at time @xmath16 $ ] with the initial condition that all walkers start from the origin @xmath26 is denoted by @xmath27 .
it was given as @xmath28 with @xmath29 for @xmath30 , where @xmath31 is the gamma function , @xmath32 and @xmath33 by using the harish - chandra / itzykson - zuber integral formula @xcite , we will see that @xmath27 is proportional to the integral @xmath34 with @xmath35 where @xmath36 is the @xmath37 diagonal matrix with @xmath38 , and the integrals @xmath39 and @xmath40 are taken over the groups of @xmath37 unitary matrices @xmath41 and real symmetric matrices @xmath42 , respectively . the proportional coefficient is determined so that the probability density is normalized . on the other hand , the integral over @xmath43 in ( [ eqn : p0nt2 ] )
can be regarded as the convolution of the gaussian distribution of complex hermitian matrices @xmath44 with variance @xmath45 and that of real symmetric matrices @xmath43 with variance @xmath46 , and thus we have the expression @xmath47 with @xmath48 where and in the following we use the abbreviations @xmath49 and @xmath50 for the real and the imaginary part of the complex number @xmath51 , respectively , i.e. @xmath52 for @xmath53 with @xmath54 .
remark that , if we set @xmath55 and @xmath56 , @xmath57 is equal to the probability density of the two - matrix model of pandey and mehta with the parameter @xmath58 @xcite .
corresponding to changing the parameter @xmath58 from 1 to 0 in the pandey - mehta two - matrix model , a gue - to - goe transition occurs in the time development of particle distribution in our vicious walks .
now we define the quantity , which we will study in the present paper ; the moment of particle positions in the vicious walks . since the distribution ( [ eqn : p0nt1 ] ) of @xmath59 is symmetric about the origin @xmath26 ,
all of the odd moments vanish .
the even moments are defined and denoted as follows , @xmath60 for @xmath61 .
first we notice that ( [ eqn : p0nt3 ] ) is invariant under any permutation of @xmath62 . then ( [ eqn : mk1 ] ) is written as @xmath63 next we introduce the integration measure for the @xmath37 complex hermitian matrices @xmath64 since @xmath65 , if @xmath66 are the eigenvalues of @xmath44 and @xmath67 ( e.g. see @xcite ) , and @xmath68 for @xmath69 with any unitary matrix @xmath70 , ( [ eqn : mk2 ] ) with ( [ eqn : mu ] ) becomes @xmath71 where @xmath72 for functions @xmath73 of the elements of @xmath44 with @xmath74 note that @xmath75 where the sum is taken over all @xmath76 combinations of indices @xmath77 , and that @xmath78 with the hermitian condition @xmath79 , the integrand @xmath80 in ( [ eqn : mk3 ] ) is a polynomial of the @xmath81 independent random variables @xmath82 . since the probability density ( [ eqn : mu2 ] ) is a product of independent gaussian integration - kernels
, we can apply the wick formula with the variances @xmath83 for @xmath84 , where @xmath85 is kronecker s delta .
we can readily prove that ( [ eqn : wick1 ] ) is equivalent with @xmath86 where @xmath87 is given by ( [ eqn : c ] ) and @xmath88 the wick formula for ( [ eqn : mk3 ] ) is thus @xmath89 with the identification @xmath90 , where the first sum is taken over all @xmath76 combinations of indices @xmath91 , and the second one over the set of permutations @xmath92 of @xmath93 with the restriction @xmath94 the total number of the terms in the second summation is @xmath95 . in this section , by performing calculation of the fourth moment @xmath96
, we will demonstrate how to obtain graphical expansions from the wick formula ( [ eqn : mk4 ] ) with the variance ( [ eqn : wick2 ] ) .
we start from the wick formula for @xmath97 , @xmath98 which has @xmath99 terms in the summand of @xmath100 .
substitution of ( [ eqn : wick2 ] ) and binomial expansion give the @xmath101 terms in the form , @xmath102 with @xmath103 ] graphically , we prepare a square with four vertices labeled @xmath104 in a cyclic order for each term as shown in fig.1 and connect the vertices @xmath105 and @xmath106 by a line for each kronecker s delta @xmath107 .
we then regard these lines connecting vertices as the hems of ribbons connecting the two edges of the square .
for example , the two lines connecting @xmath108 and @xmath109 in the term @xmath110 are considered as the two hems of a ribbon , say @xmath111 , connecting the edges @xmath112 and @xmath113 of the square , while the lines @xmath114 and @xmath115 are as those of a ribbon , say @xmath116 , connecting @xmath117 and @xmath118 .
there are two ways to connect two distinct edges by a ribbon , by untwisting as @xmath111 and by twisting as @xmath116 in the above example , respectively . for each twisted ribbon
, we put a factor @xmath119 .
next we take the summation over @xmath100 in each term . again
consider the term @xmath110 for example . under the restrictions on indices specified by the kronecker deltas , @xmath120 , only two indices ,
say @xmath121 and @xmath122 , can be chosen arbitrary from @xmath123 .
then the summation over all possible choices of indices gives @xmath81 for this term .
the contribution of @xmath110 is thus @xmath124 .
we list up the contributions of all terms in fig.1 , and the sum of them gives @xmath125 this shows that the 12 terms are classified into 6 equivalence classes , @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath130 and @xmath131 , with respect to the contribution to the moment , and fig.1 implies that all graphs for the terms in an equivalence class are topologically equivalent . for the general @xmath22-th moment ,
@xmath132 , @xmath133 , we have @xmath95 wick couplings in the formula ( [ eqn : mk4 ] ) , each term of which is the @xmath134 products of the variances @xmath135 . by applying ( [ eqn : wick2 ] ) and expanding in @xmath119
, we will have the @xmath136 terms , @xmath137 .
as demonstrated in the above section , one - to - one correspondence is established between terms @xmath138 and graphs each of which consists of a @xmath22-gon with its edges connected by @xmath134 ribbons to each other . for each graph corresponding to @xmath139 ,
let @xmath140 be the number of twisted ribbons in the @xmath134 ribbons and @xmath141 be the free indices " remaining after the identification of indices under the kronecker delta conditions .
then the contribution from the term @xmath139 is given by @xmath142 .
as mentioned at the end of the previous section , we consider the equivalence classes of the terms having the same contribution to the @xmath22-th moment , and let each equivalence class be represented by a graph @xmath143 .
we let @xmath144 and @xmath145 be the numbers of free indices and of twisted ribbons .
moreover , we denote the number of elements in the equivalence class @xmath143 by @xmath146 . in other words , @xmath146 is the number of ways to generate graphs , which are topologically equivalent with @xmath143 , using a @xmath22-gon and @xmath134 ribbons by gluing edges of @xmath22-gons by ribbons .
define @xmath147 be the collection of all graphs @xmath148 generated by the present procedure .
then we have @xmath149 each graph @xmath143 having no twisted ribbons , @xmath150 , defines a way of drawing a graph on an orientable surface @xmath151 , called a _ map _ , and each map specifies the surface @xmath151 on which the graph is drawn ( see , for example , @xcite ) . in general , the specified orientable surface has holes " or handles " and their number is called the _ genus _ @xmath152 .
the genus is related with @xmath144 , @xmath153 ( the number of distinct edges ; the original @xmath22 sides of polygon were glued together in pairs by ribbons ) and @xmath154 ( the number of faces ) through the euler characteristic , @xmath155 then the contribution from all graphs having no twisted ribbons is expressed as @xmath156 } \varepsilon_{g}(k ) \left(\frac{1}{n^2}\right)^{g } , \label{eqn : mkc0}\end{aligned}\ ] ] where @xmath157 is the indicator ; @xmath158 if the condition @xmath159 is satisfied and @xmath160 otherwise , and @xmath161 . in the similar way ,
the graphs @xmath143 having twisted ribbons , @xmath162 , are considered to define non - orientable surfaces @xmath151 .
the genus @xmath21 for non - orientable surface may be defined by the euler characteristics as @xcite @xmath163 instead of ( [ eqn : euler1 ] ) .
then all the contribution to the moment from such non - orientable surface graphs is @xmath164 where @xmath165 .
the moment ( [ eqn : mkc ] ) is then given by the summation @xmath166 it should be noted that @xmath119 defined by ( [ eqn : gamma ] ) is a monotonically increasing function of @xmath1 and changes its value from @xmath167 to 1 as the time passes from @xmath168 to @xmath2 .
the above formula ( [ eqn : mkc1 ] ) shows the fact that the contribution from mbius graphs with twisted ribbons is growing in time and at @xmath9 twisted ribbons contribute with the same weights as untwisted ribbons ( the goe case ) .
set @xmath169 for @xmath170 .
it is easy to confirm that @xmath171 is invariant under any permutation of @xmath172 and that of @xmath173 , and ( [ eqn : p0nt1 ] ) is equal to @xmath174 , if @xmath175 .
then the density function at time @xmath1 is defined as @xmath176 and the @xmath22-th moment is given by @xmath177 let @xmath178 be the @xmath14-th hermitian polynomial , satisfying the orthogonality @xmath179 with @xmath180 .
as shown in appendix a , the general formula for the dynamical correlation functions of vicious walks reported in @xcite gives the expression @xmath181 substituting ( [ eqn : r10b ] ) into ( [ eqn : mkb1 ] ) and replacing the integral variable @xmath182 by @xmath183 give @xmath184 \nonumber\\ & & + \frac{\gamma}{2^{2k+n-1}\sqrt{\pi } ( n/2 - 1 ) ! } \sum_{j=0}^{\infty } \frac{(n/2+j)!}{(n+2j+1 ) ! } \gamma^{j } \nonumber\\ & & \times \sum_{\ell=0}^{k } \frac{(2k)!}{\ell !
( 2k-2\ell ) ! }
\nonumber\\ & & \times \int h_{2k-2\ell}(y ) h_{n-1}(y ) h_{n+2j+1}(y ) e^{-y^2 } dy .
\nonumber\end{aligned}\ ] ] in the second equality , we used the christoffel - darboux formula ( see page 193 in @xcite ) @xmath185 , \nonumber \label{eqn : cd}\end{aligned}\ ] ] and the relation @xmath186 now we apply the integration formula for the triple of hermitian polynomials ( see page 290 in @xcite ) @xmath187 then we arrive at @xmath188 , with @xmath189 here we do not repeat the explanation how to characterize the quantity @xmath20 in ( [ eqn : mkc0 ] ) by using the formula ( [ eqn : mkd0 ] ) , and only mention that it is obtained as the solution of the recurrence relation @xmath190 with the boundary conditions @xmath191 see @xcite , sec.5.5 in @xcite and @xcite for details . in order to express the expansion in @xmath18 for ( [ eqn : mkd1 ] ) , here we introduce the number @xmath192 defined as the coefficients of the expansion @xmath193 for @xmath194 .
it is known that @xmath195 is the number of elements in the set @xmath196 of all permutations of @xmath197 , which are products of @xmath198 disjoint cycles .
these numbers @xmath199 are called _ the stirling numbers of the first kind _ @xcite .
for example , there are @xmath200 distinct permutations of @xmath201 .
we denote a permutation @xmath202 simply by @xmath203 $ ] @xmath204 .
we regard @xmath205 $ ] as a cycle @xmath206 , @xmath207 $ ] as a product of two cycles @xmath208 and @xmath209 , and the identity transformation @xmath210 $ ] as that of three cycles @xmath211 and @xmath209 . in this example , we see @xmath212 ( for there are two elements @xmath205 $ ] and @xmath213 $ ] with @xmath214 in @xmath215 ) , @xmath216 ( @xmath210 $ ] ) and @xmath217 ( other three permutations ) . for convenience , we will assume here that @xmath218 if @xmath219 , or @xmath220 or @xmath221
. then we have expression ( [ eqn : mkc1 ] ) from ( [ eqn : mkd1 ] ) with the coefficients @xmath222 it is easy to confirm that @xmath223 , @xmath224 and @xmath225 for any @xmath226 , and numerical tables of @xmath199 are found in @xcite .
for example , if we set @xmath227 , ( [ eqn : e ] ) gives @xmath228 , @xmath229 , @xmath230 and @xmath231 , and having @xmath232 , the result ( [ eqn : ex1 ] ) is again obtained through the general formula ( [ eqn : mkcs ] ) with ( [ eqn : mkc0 ] ) and ( [ eqn : mkc1 ] ) .
in the present paper we performed the graphical expansions for the moments of positions of vicious walkers .
the obtained formula ( [ eqn : mkcs ] ) with ( [ eqn : mkc0 ] ) and ( [ eqn : mkc1 ] ) can be regarded as the power series of the number of walkers @xmath11 .
here we consider the large @xmath11 limit .
let @xmath233 be the dominant term of ( [ eqn : mkcs ] ) in @xmath234 .
then @xmath235 that is , only the contribution from the genus - zero orientable surfaces will survives in the limit @xmath236 .
it is well known that @xmath237 s are given by the catalan numbers @xmath238 ( see , for example , @xcite ) @xmath239 and their generating function is @xmath240 corresponding to ( [ eqn : mhat ] ) , define the density function @xmath241 as @xmath242 multiplying the both sides by @xmath243 and taking the summation over @xmath134 from 0 to @xmath244 , we will have @xmath245 = \int \frac{\hat{\rho}(t , x)}{1-z x^2 } dx,\ ] ] where ( [ eqn : mhat ] ) , ( [ eqn : cat1 ] ) and ( [ eqn : cat2 ] ) were used .
if we set @xmath246 , it becomes @xmath247 this integral equation can be solved as @xcite @xmath248 in the large @xmath11 limit , the density function will keep a semicircle shape independently of the time evolution , in which only the width of the semicircle depends on time and simply scaled by @xmath87 @xcite .
in other words , _ wigner s semicircle law _ is universal in @xmath236 . the universal property of random matrix theory at the large @xmath11 limit and its finite-@xmath11 corrections
have been studied by calculating the green functions in the form of @xmath18 expansions in the field theory @xcite . in order to compare our present results with the previous ones ,
here we consider the one - point green function defined by @xmath249 with @xmath250 for @xmath53 .
it is nothing but the generating function of the moments ( [ eqn : moment0 ] ) , @xmath251 we define the coefficients @xmath252 in the expansion of the moment as @xmath253 then we have the @xmath18 expansion of the green function as @xmath254 where @xmath255 since we have obtained the closed expressions for any moments , ( [ eqn : mkcs ] ) with ( [ eqn : mkd0 ] ) and ( [ eqn : mkd1 ] ) , the green function ( [ eqn : green0 ] ) is completely determined , and the coefficients @xmath256 in its @xmath18-expansion ( [ eqn : green1 ] ) can be derived for any order @xmath257 through the formulae ( [ eqn : mkc0 ] ) , ( [ eqn : mkc1 ] ) with the known result of @xmath20 @xcite and ( [ eqn : e ] ) .
for example , as shown in appendix b , the present results give the following expressions for the first three terms ; @xmath258 and @xmath259 \nonumber\\ & & \qquad - \frac{1}{2 z \zeta } \frac{a(\zeta)}{1-a(\zeta)^2 } \left\ { 3 \gamma \frac{\partial}{\partial \gamma}-1 \right\ } \frac{\gamma a(\zeta)}{1-\gamma a(\zeta ) } \label{eqn : g2}\end{aligned}\ ] ] with @xmath260 . by using ( [ eqn :
cat2 ] ) , we can show that setting @xmath261 reduces them to equations ( 24 ) , ( 26 ) and ( 28 ) with @xmath262 ( the goe case ) given by itoi , respectively , who derived them by solving the loop equations @xcite . on the other hand , we can confirm that , if we set @xmath167 , @xmath263 gives the @xmath264 ( gue ) results of itoi .
some details are given in appendix b. our result is new and general , and it will reproduce the previous results of @xmath18-expansions for the gue and goe as the special cases with @xmath167 and 1 , respectively .
our formula also gives a power series in @xmath119 with @xmath11-dependent coefficients .
let @xmath265 .
then @xmath266 where @xmath267 are the polynomials of degree @xmath268 in @xmath11 for @xmath269 and of degree @xmath134 in @xmath11 for @xmath270
. explicit expression of @xmath271 is immediately obtained from ( [ eqn : mkd1 ] ) .
we remark that the coefficients of the highest order in @xmath119 , @xmath272 , is equal to the _ zonal polynomials _
@xmath273 , if we set all the variables @xmath274 @xcite .
as shown by ( [ eqn : p0nt3 ] ) with ( [ eqn : mu ] ) , the present system of vicious walkers can be regarded as a gaussian model as matrix model , and thus exactly solvable as demonstrated in this paper .
we would like to state , however , that the probability density ( [ eqn : p0nt1 ] ) of the positions of walkers @xmath59 is not in the simple gaussian form multiplied by @xmath275 , when @xmath276 , due to the factor @xmath277 .
combination of the pfaffian representation of this factor given in @xcite and the facts that @xmath278 for any even - dimensional antisymmetric matrix @xmath43 and @xmath279 , ( [ eqn : p0nt1 ] ) is written for even @xmath11 as @xmath280 with the non - harmonic potential @xmath281 where @xmath282 is the @xmath37 antisymmetric matrix with the element @xmath283 with @xmath284 .
then the present results for @xmath276 are nontrivial . in summary
, we demonstrated the gue - to - goe transition in time for the vicious walk model by presenting the graphical expansion formula of the moments of walker s positions .
the weights of contributing graphs are time - developing and our formula interpolates the genus expansion of orientable graphs for gue and that of non - orientable graphs for goe by using a time - parameter @xmath285 . the formula provides a power series in the number of walkers @xmath11 and it was shown that the exact expression of dynamical correlation functions recently reported by nagao
_ et al . _
@xcite is very useful in order to evaluate the coefficients in the series . by comparing with the previous results of @xmath18 expansion of the one - point green function
, we showed that our results are general and valid .
further applications of the quaternion determinantal expressions of dynamical correlations of vicious walkers to the graphical expansions of more general types of moments ( _ e.g. _ correlators of moments at different times @xcite ) will be interesting future problems .
let @xmath178 be the @xmath14-th hermite polynomial defined in sec.iv .
we can read the density function ( [ eqn : r10 ] ) from @xcite by setting @xmath286 as @xmath287 , \label{eqn : r10a}\ ] ] where @xmath288 with @xmath289 and @xmath290 the function @xmath291 can be expanded using the hermite polynomials as @xmath292 then , for @xmath293 , we will have @xmath294 where @xmath295 s satisfy the relation @xmath296 substituting ( [ eqn : r ] ) , ( [ eqn : r ] ) and ( [ eqn : phi2 ] ) into ( [ eqn : r10a ] ) gives ( [ eqn : r10b ] ) . it should be noted that , though the @xmath286 case of @xcite is equivalent with pandey - mehta s two - matrix model @xcite , the present expression ( [ eqn : r10b ] ) is more useful for the moment calculation as shown in sec.iv .
by definition of @xmath252 in ( [ eqn : mbar ] ) , ( [ eqn : mkc0 ] ) and ( [ eqn : mkc1 ] ) give @xmath297 where we have used ( [ eqn : e ] ) , ( [ eqn : cat1 ] ) , and the fact @xmath298 , which is obtained from ( [ eqn : rec ] ) with ( [ eqn : cat1 ] ) . through (
[ eqn : cat2 ] ) and ( [ eqn : gtn ] ) , ( [ eqn : mbar0 ] ) immediately gives ( [ eqn : g0 ] ) . in order to derive ( [ eqn : g1 ] ) and ( [ eqn : g2 ] ) from ( [ eqn : mbar1 ] ) and ( [ eqn : mbar2 ] ) , respectively
, we can use the identity given as equation ( c.5 ) in @xcite , @xmath299 and its derivatives with respect to @xmath300 .
it is easy to see that @xmath301 and @xmath302 is zero at @xmath167 .
moreover , we have obtained the following expressions for @xmath303 @xmath304 -\frac{1}{2(z^2 - 2)^{3/2 } } ( 1-\gamma)+{\cal o}((1-\gamma)^2 ) , \nonumber\\ & & g_{t}^{(2)}(t , z)= \frac{2z^2 + 3 - 2z \sqrt{z^2 - 2 } } { 4 ( z^2 - 2)^{5/2 } } -\frac{z(z^2 + 6)-(z^2 + 1)\sqrt{z^2 - 2}}{4(z^2 - 2)^3}(1-\gamma ) + { \cal o}((1-\gamma)^2 ) , \nonumber\end{aligned}\ ] ] the authors thank s. kakei and t. sasamoto for letting them know the papers by m. mulase .
they also thank t. nagao for useful comments on the present work .
they acknowledge the comments of a referee , which enabled them to show the calculations of one - point green function in the present paper . | a system of brownian motions in one - dimension all started from the origin and conditioned never to collide with each other in a given finite time - interval @xmath0 $ ] is studied .
the spatial distribution of such vicious walkers can be described by using the repulsive eigenvalue - statistics of random hermitian matrices and it was shown that the present vicious walker model exhibits a transition from the gaussian unitary ensemble ( gue ) statistics to the gaussian orthogonal ensemble ( goe ) statistics as the time @xmath1 is going on from 0 to @xmath2 . in the present paper , we characterize this gue - to - goe transition by presenting the graphical expansion formula for the moments of positions of vicious walkers . in the gue limit @xmath3 ,
only the ribbon graphs contribute and the problem is reduced to the classification of orientable surfaces by genus . following the time evolution of the vicious walkers ,
however , the graphs with twisted ribbons , called mbius graphs , increase their contribution to our expansion formula , and we have to deal with the topology of non - orientable surfaces . application of the recent exact result of dynamical correlation functions yields closed expressions for the coefficients in the mbius expansion using the stirling numbers of the first kind . |
isospin symmetry was introduced in the thirties by heisenberg in his studies of the atomic nucleus . since then , many particles have been found to appear in iso multiplets , like the nucleons , the pions , the delta isobars , a.s.o . with the advent of qcd ,
a deeper understanding of isospin symmetry has emerged . in the limit of equal up and down current quark masses and in the absence of electroweak interactions
, isospin is an exact symmetry of qcd .
the intra
multiplet mass splittings allow to quantify the breaking of this symmetry , which is caused by different mechanisms ( for a detailed review , see ref.@xcite ) .
first , the light quark masses are everything but equal ( still , their absolute masses are much smaller than any other qcd scale and thus this breaking can be treated as a perturbation ) .
second , the light quarks have different charges and thus react differently to the electromagnetic ( em ) interactions . the em effects are also small since they are proportional to the fine structure constant @xmath0 . in the case of the pions ,
the mass splitting is almost entirely of em origin .
this can be traced back to the absence of d like couplings in su(2 ) , thus promoting the quark mass difference to a second order effect .
for the nucleons , matters are different , strong and em effects are of similar size but different signs . the fact that the neutron is heavier than the proton leads to the conclusion that @xmath1 , consistent with the analysis of the kaon masses . since we know that isospin is broken - so why bother ?
first , the picture that has emerged from the hadron masses can not be considered complete , there is still on going discussion about the size of the violation of dashen s theorem , the possibility of a vanishing up quark mass to solve the strong cp problem and `` strange '' results from lattice gauge theory .
also , the analysis of the quark mass dependence of the baryon masses remains to be improved ( for a classic , see ref.@xcite and a recent study , see ref.@xcite ) .
furthermore , only a few dynamical implications of isospin violation have been verified experimentally and a truely quantitative picture has not yet emerged .
in addition , the nucleus as a many body system offers a novel laboratory to study isospin violation .
in addition , with the advent of cw electron accelerators and improved detectors , we now have experimental tools to measure threshold pion photoproduction with an unprecedented accuracy .
the purely mesonic sector was not touched upon in this working group , but there is one recent result which i would like to discuss . in elastic @xmath2 scattering ,
the chiral perturbation analysis has been carried out to two loops .
it was demonstrated in refs.@xcite that the em isospin violating effects are of the same size as the hadronic two loop corrections . for a precise description of low energy pion reactions , it is thus mandatory to include such effects consistently .
a somewhat surprising result was found in case of the scalar and the vector form factor of the pion in ref.@xcite .
it was shown that the em corrections to the momentum dependence of both form factors are tiny ( due to large cancellations between various contributions ) , much smaller than the corresponding hadronic two loop contributions worked out in refs.@xcite .
this result remains to be understood in more detail .
it is particularly surprising for the scalar form factor since it is not protected by a conserved current theorem la ademello gatto .
only the normalization of the scalar form factor exhibits the few percent em corrections anticipated from the study of the @xmath2 scattering lengths .
note , however , that the smallness of the effects of the light quark mass difference for the pion form factors has been known and understood since long @xcite .
the pion nucleon system plays a particular role in the study of isospin violation .
first , the explicit chiral symmetry breaking and isospin breaking operators appear at the same order in the effective lagrangian which maps out the symmetry breaking part of the qcd hamiltonian , i.e. the quark mass term ( restricted here to the two lightest flavors ) , @xmath3 so that the strong isospin violation is entirely due to the isovector term whereas the isoscalar term leads to the explicit chiral symmetry breaking . in the presence of nucleons ( and in contrast to the pion case ) , both breakings appear at the same order .
this can lead to sizeable isospin violation as first stressed in reactions involving neutral pions by weinberg @xcite .
let me perform some naive dimensional analysis for the general case ( say for any given channel in @xmath4 scattering that is not suppressed to leading order ) .
isospin violation ( iv ) should be of the size @xmath5 where the mass of the @xmath6 set the scale for the non goldstone physcis . in the presence of a close by and strongly coupled baryonic resonance like the @xmath7
, iv might be enhanced @xmath8 of course , such type of arguments can not substitute for full scale calculations .
second , there are two analyses @xcite which seem to indicate a fair amount of isospin violation ( of the order of 6 ... 7% , which is much bigger than the dimensional arguments given above would indicate ) in low
energy @xmath4 scattering , see gibbs talk @xcite
. this can not be explained in conventional meson
exchange models by standard meson mixing mechanisms .
i would also like to mention that in these two analyses the hadronic and the electromagnetic contributions are derived from different models .
this might cause some concern about possible uncertainties due to a theoretical mismatch . clearly , it would be preferable to use here one unique framework . that can ,
in principle , be supplied by chiral perturbation theory since electromagnetic corrections can be included systematically by a straightforward extension of the power counting .
this is most economically , done by counting the electric charge as a small parameter , i.e. on the same footing as the external momenta and meson masses . the heavy baryon chiral perturbation theory machinery to study these questions to complete one loop ( fourth )
order has been set up as shown by mller @xcite .
it is important to perform such calculations to fourth order since one loop graphs appear at dimension three _ and _ four .
furthermore , it is known from many studies that one loop diagrams with exactly one insertion from the dimension two @xmath4 lagrangian are ( often ) important .
finally , symmetry breaking ( chiral and isospin ) in the loops only starts at fourth order .
in particular , questions surrounding the @xmath9 @xmath10term or neutral pion scattering off nucleons can now be addressed to sufficient theoretical precision . a first step in this direction for all channels in @xmath4 scattering
was reported by fettes @xcite , but a full scale one loop calculation including all virtual photon effects still has to be done . of particular interest is the novel relation between @xmath11 and @xmath12 scattering off protons that is extremely sensitive to isospin violation .
it should also be stressed that for such tests , it is mandatory to better measure and determine the small isoscalar @xmath9 amplitudes . also , the relations which include the much bigger isovector amplitudes show iv consistent with the dimensional arguments given in eq.([dim ] ) .
i consider the `` ordering schemes '' discussed by gibbs and fettes very useful tools to pin down the strengths and sources of isospin breaking in @xmath4 scattering .
this also allows to see a priori which type of measurements are necessary to obtain complete information and to what extent various reactions can give redundant information ( one example is discussed by gibbs @xcite ) .
intimately related to this is pion photoproduction via the final state theorem , i.e. certain @xmath4 scattering phases appear in the imaginary part of the respective charged or neutral pion photoproduction multipoles .
bernstein @xcite stressed that in neutral pion photoproduction off protons , there are two places to look for isospin violation .
one is below the @xmath13 threshold , which might give access to the elusive ( but important ) @xmath14 scattering length . at present
, it does not appear that the original proposal of measuring the target polarization below the @xmath13 threshold to high precision is feasible at a machine like e.g. mami .
the other important effect , which appears to be more easily accessible to an experiment , is the strength of the cusp at the opening of the @xmath13 threshold , which according to bernstein s three
channlel s
matrix analysis @xcite is quite sensitive to isospin violation .
such a calculation should also be done in the framework of heavy baryon chiral perturbation theory ( beyond the charged to neutral pion mass difference effects included so far ) . over the last years
, there has been a very fruitful interplay between experimenters and theorists particularly in the field of pion photo and electroproduction and it is of utmost important to further strengthen this . it is a theorists dream that reactions with neutral pions ( elastic scattering and photoproduction ) will be measured to a high precision .
an important point was stressed by lewis @xcite . in a `` toy '' calculation ( i.e. an su(2 ) approach to the strange vector form factor of the nucleon , which is clearly related to three flavor qcd )
, he showed that isospin breaking effects can simulate a `` strange '' form factor that intrinsically vanishes in that approach .
this nicely demonstrates that to reliably determine small quantities , may they be related to isospin conserving or violating operators , _ all _ possible effects have to be included .
the recent measurements at bates and jlab , which seem to indicate small expectation values of the strange vector current in the proton , should therefore be reanalyzed . in this case
, isospin violation appears to be a nuisance but can not be ignored .
the only new data with respect to iv were presented by machner @xcite .
he analysed recent data from cosy and iucf for @xmath15 and @xmath16 . for exact isospin symmetry ( i.e. after removing the coulomb corrections )
, the pertinent cross sections should be equal ( up to a clebsch ) . in the threshold region , one can make a partial wave expansion and finds that the s wave contribution @xmath17 shows iv of the order of 10% and no effect is observed in the p wave terms . to my knowledge , a theoretical understanding of this effect is lacking . despite a huge amount of efforts over the last years , a model
independent effective field theory description of pion production in proton proton collisions has not yet been obtained .
the energies involved to even produce a pion at rest are too large for the methods employed so far .
more progress , however , has been made in the two nucleon system at low energies .
it is well known that iv appears in the nn scattering lengths . in the nuclear jargon ,
one talks about charge independence breaking ( cib ) ( @xmath18 after coulomb subtraction , where @xmath19 denotes the scattering length ) and charge symmetry breaking ( csb ) ( @xmath20 after coulomb subtraction ) .
these effects are naturally most pronounced at threshold .
kaplan , savage and wise ( ksw ) @xcite have proposed a non perturbative scheme that allows for power counting on the level of the nucleon
nucleon scattering amplitude . in that framework ,
iv ( cib and csb ) has recently been investigated @xcite .
it was shown that isospin violation can be systematically included in the effective field theory approach to the two nucleon system in the ksw formulation .
for that , one has to construct the most general effective lagrangian containing virtual photons and extend the power counting accordingly .
this framework allows one to systematically classify the various contributions to cib and csb .
in particular , the power counting combined with dimensional analysis allows one to understand the suppression of contributions from a possible charge
dependence in the pion
nucleon coupling constants . including the pions , the leading cib breaking effects are the pion mass difference in one pion exchange together with a four nucleon contact term .
these effects scale as @xmath21 , where @xmath22 is the genuine expansion parameter of the ksw scheme .
power counting lets one expect that the much debated contributions from two pion exchange and @xmath23 graphs are suppressed by factors of @xmath24 and @xmath25 , respectively .
this is in agreement with some , but not all , previous more model dependent calculations . the leading charge symmetry breaking is simply given by a four nucleon contact term .
often , the nucleus can be used as a filter to enhance or suppress certain features of reactions as they appear in free space .
furthermore , measurements on the neutron , which are necessary to get the complete information in the isospin basis ( for a discussion on this topic with respect to pion photoproduction , see ref.@xcite ) can only be done on ( preferably polarized ) light nuclei .
gibbs @xcite has pointed out that a measurement of charge exchange on the proton and the neutron ( in forward direction and close to the interference minimum near 45 mev ) could be done in the @xmath26he
triton system .
this would be an interesting possibility to get another handle on the elusive neutron and allow one to pin down one of the amplitudes parametrizing iv ( according to the ordering scheme mentioned above ) . for a more detailed discussion concerning the extraction of neutron properties from the deuteron
, i refer to the recent summary by beane @xcite .
for sure , isospin symmetry is broken .
however , do we precisely know the size of iv from experiment ?
the answer is yes and no .
we have some indicative information but no systematic investigations of all pertinent low energy reactions are available . also , one might ask the question whether the methodology , which has been used so far to extract numbers on iv , say from low energy @xmath4 scattering data , is reliable ? if we assume that this is the case , we still have no deeper understanding of the mechanisms triggering iv .
to my knowledge , the only machinery to consistently separate strong and electromagnetic iv is based on effective field theory . in that scheme
, one can consider various reactions like elastic @xmath4 scattering , pion photoproduction or even nucleon compton scattering to try to get a handle on the symmetry breaking operators @xmath27 .
also , a systematic treatment of isospin violation is mandatory for the determiantion of small quantities like the isoscalar s wave scattering length or the strange nucleon form factors .
based on that , i have the following wish list for theory and experiment : theory : 1 .
the effective chiral lagrangian calculations can and need to be improved .
in particular , it is most urgent to get a handle on the so called low energy constants , which parametrize the effective lagrangian beyond leading order .
sum rules , models or even the lattice might be useful here .
2 . a deeper theoretical understanding of certain phenomenological models ( like e.g. the extended tree level model of ref.@xcite ) in connection with the approaches to correct for coulomb effects would be helpful .
3 . the dispersion theoretical approach should be revisited and set up in a way to properly include iv ( beyond what has been done so far ) .
for some first steps , see the talk by oades @xcite .
experiment : 1 . clearly , we need more high precision data for the elementary processes , but not only for @xmath4 scattering but also for ( neutral ) pion photo / electroproduction .
more precise nuclear data are also needed .
embedding the elementary reactions in the nucleus as a filter allows one to get information on the elusive neutron properties .
clearly , this refers to few
nucleon systems where precise theoretical calculations are possible .
finally , i would like to stress again that a truely quantitative understanding of isospin violation can only be obtained by considering a huge variety of processes . while pion
nucleon scattering is at the heart of these investigations , threshold pion photoproduction or the nucleon form factors also play a vital role in supplying additional information .
i would like to thank all participants of this working group for their contributions .
i am grateful to my collaborators evgeny epelbaum , nadia fettes , bastian kubis , guido mller and sven steininger for sharing with me their insight into this topic .
last but not least the superbe organization by christine kunz and res badertscher is warmly acknowledged .
j. gasser and h. leutwyler , phys . rep . * c87 * , 77 ( 1982 ) . j. gasser , ann . phys .
( ny ) * 136 * , 62 ( 1981 ) .
b. borasoy and ulf - g .
meiner , ann .
( ny ) * 254 * , 192 ( 1997 ) .
meiner , g. mller , and s. steininger , phys . lett . *
b406 * , 154 ( 1997 ) 154 ; ( e ) _ ibid _ * b407 * , 434 ( 1997 ) . m. knecht and r. urech , nucl .
b519 * , 329 ( 1998 ) .
b. kubis and ulf - g .
meiner , hep - ph/9908261 .
j. gasser and ulf - g .
meiner , nucl . phys .
* b357 * , 90 ( 1991 ) . j. bijnens , g. colangelo , and p. talavera , jhep * 9805 * , 014 ( 1998 ) . j. gasser and h. leutwyler , ann .
phys.(ny ) * 158 * , 142 ( 1984 ) .
s. weinberg , trans .
n.y . acad . of sci . * 38 * , 185 ( 1977 ) .
gibbs , li ai and w.b .
kaufmann , phys .
lett . * 74 * , 3740 ( 1995 ) .
e. matsinos , phys . rev . *
c56 * , 3014 ( 1997 ) .
gibbs , _ these proceedings_. g. mller , _ these proceedings_. n. fettes , _
these proceedings_. a.m. bernstein , _ these proceedings_. r. lewis , _ these proceedings_. h. machner , _ these proceedings_. d.b .
kaplan , m.j .
savage and m.b .
wise , nucl .
b534 * , 329 ( 1998 ) .
e. epelbaum and ulf - g .
meiner , phys . lett . *
b467 * , 284 ( 1999 ) .
meiner , nucl . phys . *
a629 * , 72c ( 1998 ) .
beane , nucl - th/9909016 . p.f.a .
goudsmit et al . , nucl . phys . * a575 * , 673 ( 1994 ) .
g. oades , _ these proceedings_. | i give an introduction to the problem of isospin violation and add some comments to the various topics addressed in the working group . |
the orientational order of liquid crystals can be controlled by electric or magnetic fields due to the anisotropy of dielectric permittivity and diamagnetic susceptibility . in a restricted geometry ( between parallel substrates )
this effect is called the fredericksz transition . the classical fredericksz transition in pure nematics has been studied in detail both experimentally and theoretically @xcite .
the relatively large anisotropy of the dielectric permittivity in usual nematics triggers the orientational response to an electric voltage already at a few volts .
in contrast , because of the small value of the anisotropy of the diamagnetic susceptibility ( @xmath6 @xmath7 10@xmath8 ) , magnetic fields @xmath9 used for the same purpose have to reach rather large values ( @xmath10 t ) . in an effort to enhance the magnetic susceptibility of liquid crystals , the idea of doping them with fine magnetic particles was theoretically introduced by brochard and de gennes .
they constructed a continuum theory of magnetic suspensions in nematic liquid crystals ( ferronematics ) in their fundamental paper @xcite , prior to the chemical synthesis of these systems .
many reports have shown that doping of liquid crystals with small amount of magnetic nanoparticles can lead to the decrease as well as to the increase of the threshold @xmath11 of the magnetic fredericksz transition @xcite , depending on the anisotropy of diamagnetic susceptibility @xmath6 of the nematic host , and on the initial mutual orientation of the nematic director @xmath12 and the magnetic moment @xmath13 of the magnetic particles .
for example , when @xmath14 and @xmath15 , doping with magnetic nanoparticles decreases @xmath16 @xcite , while in case of combination @xmath17 and @xmath18 @xcite , or that of @xmath14 and @xmath19 @xcite , doping with the nanoparticles increases @xmath16 . in recent works by podoliak et al .
@xcite , and buluy et al .
@xcite both experimental and theoretical investigations have been reported about the _ optical response _ of suspensions of ferromagnetic nanoparticles in nematic liquid crystals on the imposed magnetic field ( which finally leads to a fredericksz transition ) .
the authors have measured an additional , linear response in ferronematics at low magnetic fields ( far below @xmath16 ) . in their analysis a theoretical model based on the burylov - raikher theory @xcite has been presented .
we have to note here that in refs .
@xcite a weak bias magnetic field ( @xmath20 mt ) has also been applied parallel to @xmath12 in order to align the magnetic dipole moments of the nanoparticles parallel with @xmath12 as an initial condition .
these recent works have inspired us to get more insights by approaching the phenomenon from a different experimental aspect : to perform _ capacitance measurements _ on the nematic liquid crystal 6chbt doped with spherical magnetic nanoparticles , single - wall carbon nanotubes ( swcnt ) , swcnt functionalized with carboxyl groups ( swcnt - cooh ) and swcnt functionalized with fe@xmath0o@xmath1 nanoparticles ( swcnt / fe@xmath0o@xmath1 ) , however , without an aligning bias magnetic field ( @xmath21 ) .
the synthesis of the spherical magnetic nanoparticles was based on co - precipitation of fe@xmath22 and fe@xmath23 salts by nh@xmath1oh at 60 @xmath24c @xcite . the size and morphology of the particles
were determined by transmission electron microscopy ( tem ) .
the mean diameter of the obtained spherical magnetic nanoparticles was 11.6 nm .
swcnt and swcnt - cooh ( produced by catalytic chemical vapor deposition technique ) were purchased from cheap tubes inc . , and then functionalized with fe@xmath0o@xmath1 nanoparticles .
the synthetic route of the functionalization is described in details in ref .
@xcite , and the final swcnt / fe@xmath0o@xmath1 product is shown schematically in fig .
[ sample ] .
swcnts ( of mixed chirality ) exhibit in about 60@xmath25 semiconducting property and in about 40@xmath25 metallic one .
the length of nanotubes ranged from 0.5 @xmath26 m to 2 @xmath26 m , their outer diameter from 1 nm to 2 nm and inner diameter : 0.8 nm - 1.6 nm .
the magnetic properties of all prepared nanoparticles were investigated by squid magnetometer ( quantum design mpms 5xl ) .
the ferronematic samples were based on the thermotropic nematic 4-(trans-4-n - hexylcyclohexyl)-isothiocyanatobenzene ( 6chbt ) , which was synthetized and purified at the institute of chemistry , military technical university , warsaw , poland .
6chbt is a low - temperature - melting enantiotropic liquid crystal with high chemical stability @xcite .
the phase transition temperature from the isotropic liquid to the nematic phase has been found to be 42.6 @xmath24c .
the heptan - based dispersion of spherical magnetic nanoparticles was added to the liquid crystal in the isotropic state under stirring until the solvent has evaporated .
the doping with nanotubes was done by adding particles under continuous stirring to the liquid crystal in the isotropic phase .
the volume concentrations of the nanoparticles were @xmath2 = 10@xmath3 and @xmath4 = 10@xmath5 , i.e. , small enough to keep the interparticle magnetic dipole - dipole interaction ignorably small @xcite .
structural transitions in ferronematic samples have been monitored by capacitance measurements in a capacitor made of indium - tin - oxide ( ito ) coated glass electrodes .
the capacitor with the electrode area of approximately ( @xmath27 ) @xmath28 has been connected to a regulated thermostat system , with a temperature stability of 0.05 @xmath24c .
measurements were performed at the temperature of 35 @xmath24c .
the distance between the electrodes ( the sample thickness ) was @xmath29 @xmath26 m
. the capacitance was measured at the frequency of 1 khz and voltage of 0.1 v by a high precision capacitance bridge andeen hagerling . in the experiments , the liquid crystal had planar alignment .
the planar alignment was achieved by polyimide layers rubbed antiparallel , which produces a small pre - tilt @xmath30 typically between @xmath31 and @xmath32 see e.g. , in ref .
the capacitance of the cells had a value @xmath33 for this alignment .
the magnetic field was applied perpendicular to the electrode surfaces ( as shown in fig .
[ schema ] ) .
the measurements were done immediately after the cells were filled .
in the pure 6chbt the magnetic fredericksz transition starts at 2.63 t. due to doping , the fredericksz threshold is shifted to lower values @xcite , but it is still higher than 2 t. figures [ spher ] , [ swcnt ] , [ swcnt - cooh ] and [ swcnt - fe ] show the relative capacitance variation @xmath34 of the 6chbt liquid crystal doped with spherical magnetic nanoparticles , swcnt , swcnt - cooh and swcnt / fe@xmath0o@xmath1 , respectively , as a function of the magnetic induction @xmath35 in the low magnetic field range ( up to 0.1 t ) , far below the threshold of the magnetic fredericksz transition .
the figures provide a clear evidence for a linear magnetic field dependence of the _ capacitance _ in this low magnetic field region , i.e. for the presence of an effect similar to that obtained by an _
optical _ method as reported in refs .
the magnetic field dependence of @xmath36 is stronger for a higher volume concentration of nanoparticles in all cases , but it is not present in the undoped liquid crystal .
the sensitivity of the suspensions to the field depends on the type of the nanoparticle used .
the lowest response was detected for doping with swcnt - cooh , while a much larger response was measured for doping with spherical magnetic particles fe@xmath0o@xmath1 and swcnt / fe@xmath0o@xmath1 .
even the nematic suspension with the net swcnt showed a considerably larger effect than that with swcnt - cooh .
note that for the suspension with the larger swcnt / fe@xmath0o@xmath1 concentration ( @xmath37 ) a different ( nonlinear ) magnetic field dependence of the reduced capacitance is detected .
a qualitatively similar low magnetic field effect has been measured optically in nematic colloids of carbon nanotubes filled with @xmath38-fe , in which a strong aggregation has been observed on the time scale of hours [ 9 ] .
the data shown above confirm that ferronematic suspensions may show well measurable response in capacitance to the magnetic field even much below the classical magnetic fredericksz threshold .
our experimental results , on the one hand , are in agreement with the recent reports @xcite , regarding the dependence of the effect on the concentration and type of nanoparticles . on the other hand ,
our measurements indicate that the aligning bias magnetic field ( used in refs .
@xcite ) is not an essential prerequisite for the nematic suspensions to be sensitive to low magnetic fields .
the possible reason for this deviation regarding b@xmath39 is that the ferronematics were based on different nematic hosts in these experiments . in ref .
@xcite the 4-cyano-(4-pentyl)biphenyl ( 5cb ) has been used , while in ref .
@xcite the mixture e7 served as a host , which is also composed of various cyanobiphenyls . for the compound 8cb ( which belongs to the same family ) it
was shown @xcite that it prefers a perpendicular initial alignment ( @xmath40 ) between the director @xmath41 and the magnetic moment @xmath42 of the nanoparticle .
the role of the bias magnetic field could be to overwrite this initial condition yielding @xmath43 . in the 6chbt - based ferronematics used by us
, however , @xmath44 is the preferred initial condition ( even in the absence of a bias field ) @xcite .
a much more challenging task is to interpret the detected linear response in capacitance to low magnetic fields .
the continuum theory allows the possibility for such a linear coupling only in the presence of both the non - zero initial pre - tilt @xmath30 , and the magnetic dipole moment @xmath42 of the nanoparticles .
namely , without pre - tilt , in the limit of small changes , the dielectric permittivity ( and by that , the capacitance ) depends only quadratically on the change in the director out - of - plane tilt @xmath45
. however , when a pre - tilt is present , a linear term @xmath46 will also enter the expression for the permittivity . on the other hand
, the free energy of the system ( which has to be minimized to obtain a solution ) will contain a term linear on the magnetic field only if the magnetic moment of the magnetic particles @xmath42 is taken into account through the dipolar interaction between the particles and the magnetic field
see eq . 2 of ref .
the proper combination of these effects of @xmath30 and @xmath42 may eventually lead to a solution showing the experimentally observed linear c(b ) dependence .
therefore , in this regard further theoretical and experimental investigations are desirable , primarily concentrating on the role of the pre - tilt @xmath30 .
theoretically the introduction of @xmath30 into the burylov - raikher theory , while experimentally a systematic variation of the pre - tilt ( by different surface treatments ) could serve as the first step forward .
interestingly , 6chbt doped with `` non - magnetic '' swcnt ( fig .
[ swcnt ] ) has a comparable low magnetic field response to those ferronematics obtained by doping 6chbt with magnetic particles ( spherical fig .
[ spher ] , and swcnt / fe@xmath0o@xmath1 fig .
[ swcnt - fe ] ) , while in 6chbt doped with swcnt - cooh the detected response is considerably lower ( though still non - zero see fig .
[ swcnt - cooh ] ) .
these observations lead to two main conclusions .
firstly , both `` non - magnetic '' swcnts and swcnt - cooh particles have a non - zero magnetization .
this has actually been measured for our swcnt @xcite , and is also supported by both theoretical and experimental investigations performed on carbon nanotubes , which have shown that nanotubes may have diamagnetic as well as paramagnetic properties depending primarily on their electronic structure ( see e.g. , @xcite ) . secondly , the functionalization ( either with cooh or with fe@xmath0o@xmath1 ) of the carbon nanotubes ( which are of mixed chirality , and exhibit both semiconducting and metallic properties ) leads to further complications .
namely , the functionalization alters also the interaction between the particles and the liquid crystal ( i.e. , the anchoring energies are different ) .
therefore , magnetic properties of the given swcnt , as well as the type of the anchoring , and the magnitude of the anchoring energy has to be taken into account to interpret these results in a proper way . the results and conclusions presented above imply that a deeper , more detailed analysis of the theoretical model @xcite , focusing on the low magnetic field limit is necessary for the interpretation of the low field response ( both optical and capacitive ) of ferronematic suspensions .
this statement is also supported by the optical results of ref .
@xcite for higher concentrations ( @xmath47 ) of nanoparticles , where only a moderate agreement between the model and the experimental curves has been achieved .
the work was supported by the slovak academy of sciences , in the framework of cex - nanofluid , projects vega 0077 , the slovak research and development agency under the contract no .
apvv-0171 - 10 , ministry of education agency for structural funds of eu in frame of project 6220120021 , the grenoble high magnetic field laboratory ( creta ) and the hungarian research fund otka k81250 .
n.e . is grateful for the hospitality and support provided by the ministry of education agency for structural funds of eu in the frame of project 26110230061 .
de gennes and j. prost , _ the physics of liquid crystals _ ( clarendon press , oxford , 1993 )
. f. brochard and p.g .
de gennes , j. phys .
( paris ) * 31 * , 691 ( 1970 ) .
n. tomaoviov , m. konerack , p. kopansk , m. timko , v. zviov , and j. jadzyn , phase transitions * 79 * , 595 ( 2006 ) .
p. kopansk , n. tomaoviov , m. konerack , v. zviov , m. timko , a. darov , a. princov , n. ber , k. fodor - csorba , t. tth - katona , a. vajda , and j. jadzyn , phys rev e * 78 * , 011702 ( 2008 ) .
e. petrescu and e.r .
bena , j. magn .
magn . mater . *
321 * , 2757 ( 2009 ) .
martinez - miranda and l.k .
kurihara , j. appl .
phys . * 105 * , 084305 ( 2009 ) .
p. kopansk , n. tomaoviov , m. konerack , m. timko , v. zviov , n. ber , k. fodor - csorba , t. tth - katona , a. vajda , j. jadzyn , e. beaugnon , and x. chaud , j. magn .
magn . mater . *
322 * , 3696 ( 2010 ) .
p. kopansk , m. konerack , i. potoov , m. timko , l. tomo , j. jadzyn , and g. czechovski , czech . j. phys .
* 51 * , 59 ( 2001 ) .
n. podoliak , o. buchnev , o. buluy , g. dalessandro , m. kaczmarek , y. reznikov , and t.j .
sluckin , soft matter * 7 * , 4742 ( 2011 ) .
o. buluy , s. nepijko , v. reshetnyak , e. ouskova , v. zadorozhnii , a. leonhardt , m. ritschel , g. schnhense , and y. reznikov , soft matter * 7 * , 644 ( 2011 ) .
s.v . burylov and y.l .
raikher , phys .
e * 50 * , 358 ( 1994 ) .
s.v . burylov and y.l .
raikher , mol .
technol . , sect .
a * 258 * , 107 ( 1995 ) ; _ ibid . _
* 258 * 123 ( 1995 ) .
z. mitrov , n. tomaoviov , m. timko , m. konerack , j. kov , j. jadzyn , i. vvra , n. ber , t. tth - katona , e. beaugnon , x. chaud , and p. kopansk , new j. chem .
* 35 * , 1260 ( 2011 ) .
r. dabrowski , j. dziaduszek , and t. szczucinski , mol .
. lett . * 102 * , 155 ( 1984 ) .
h. qi , b. kenkead , and t. hegmann , proc .
spie * 6911 * , 691106 ( 2008 ) .
p. kopansk , i. potoov , m. konerack , m. timko , a.g.m .
jansen , j. jadzyn , and g. czechowski , j. magn .
magn . mater . *
289 * , 101 ( 2005 ) .
tsai , s.c .
chen , f.l .
shyu , c.p .
chang , and m.f .
lin , physica e * 30 * , 86 ( 2005 ) .
shyu , c.p .
chang , r.b .
chen , c.w .
chiu , and m.f .
lin , phys .
b * 67 * , 045405 ( 2003 ) .
lin and k.w .- k .
shung , phys .
b * 52 * , 8423 ( 1995 ) .
s. glenis , v. likodimos , n. guskos , and c.h .
lin , j. magn .
magn . mater . * 272 - 276 * , 1660 ( 2004 ) .
j. heremans , c.h .
olk , and d.t .
morelli phys .
b * 49 * , 15122 ( 1994 ) . o@xmath1 nanoparticles .
the inset shows the magnetic fredericksz transition for the sample doped with magnetic particles of volume concentration @xmath48 .
symbols are the experimental data , and the lines represent linear fit to them . ] | the response in capacitance to low external magnetic fields ( up to 0.1 t ) of suspensions of spherical magnetic nanoparticles , single - wall carbon nanotubes ( swcnt ) , swcnt functionalized with carboxyl group ( swcnt - cooh ) and swcnt functionalized with fe@xmath0o@xmath1 nanoparticles in a nematic liquid crystal has been studied experimentally .
the volume concentration of nanoparticles was @xmath2 = 10@xmath3 and @xmath4 = 10@xmath5 .
independent of the type and the volume concentration of the nanoparticles , a linear response to low magnetic fields ( far below the magnetic frederiksz transition threshold ) has been observed , which is not present in the undoped nematic . |
the ursa minor dwarf spheroidal galaxy ( dsph ) , detected by wilson ( 1955 ) on plates of the palomar sky survey , is at @xmath11 ( j2000.0 ; mateo 1998 ) on the sky .
its galactic coordinates are @xmath12 .
several published estimates of the distance to ursa minor disagree by more than their combined uncertainties .
olszewski & aaronson ( 1985 ) and cudworth , olszewski , & schommer ( 1986 ) determine a distance modulus , @xmath13 , of @xmath14 , which corresponds to a heliocentric distance of @xmath15 kpc .
these two studies are not independent because they use the same photometric calibration .
this distance is in agreement with those measured by nemec , wehlau , & mendes de oliveira ( 1988 ) , @xmath16 kpc , and mighell & burke ( 1999 ) , @xmath17 kpc .
the latter value comes from @xmath18 and @xmath19 images of the central region in ursa minor taken with the _ hubble space telescope _ ( hst ) .
however , carrera ( 2002 ) determine @xmath20 , which gives a distance of @xmath21 kpc ( note an erroneous entry of 70 kpc for the distance in their table 3 ) .
bellazzini ( 2002 ) find a similar value : @xmath22 .
about 0.3 mag of the 0.41 mag difference between the distance moduli of cudworth , olszewski , & schommer ( 1986 ) and bellazzini ( 2002 ) comes from different values for the absolute magnitude of the horizontal branch .
it is beyond the scope of this article to resolve the uncertainties in the globular cluster distance scale , however the larger distance is based on more modern stellar models and is probably in better agreement with hipparcos parallaxes for subdwarfs ( reid 2003 ; gratton 2003 ) and the age of the universe derived by wmap ( bennett 2003 ) .
thus , this study adopts a distance of 76 kpc for the purpose of deriving distance - dependent quantities .
ursa minor is one of the least luminous galactic dsphs .
irwin & hatzidimitriou ( 1995 : ih hereafter ) report its luminosity to be @xmath23 .
reflecting their larger estimate of the distance , carrera ( 2002 ) report @xmath24 .
these luminosities are based on the structural parameters from ih : a core radius of @xmath25 arcmin and a tidal radius of @xmath26 arcmin .
kleyna ( 1998 ) find similar structural parameters for ursa minor .
however , martnez - delgado ( 2001 ) and palma ( 2003 ) find a more extended radial profile , which implies a larger luminosity .
palma ( 2003 ) determine that both blue horizontal branch stars and stars identified as giants on the basis of photometry in the washington - band system extend beyond the tidal radius of ih .
they argue that the luminosity of ursa minor is 2.7@xmath27 larger than the value of ih .
based on both the larger radial extent and the larger distance , bellazzini ( 2002 ) adopt a luminosity of @xmath28 .
our study likewise adopts this value .
several studies find that the stars of ursa minor are old , metal - poor , and of similar age .
olszewski & aaronson ( 1985 ) constructed a @xmath18 _ versus _
@xmath29 color - magnitude diagram for the dsph using data from ground - based observations .
the diagram reveals that the stars of ursa minor have ages and metallicities similar to those of the metal - poor globular cluster m92 .
the study interprets the few stars brighter and bluer than the main - sequence turnoff as blue stragglers .
carrera ( 2002 ) confirm this basic picture using ground - based data in the @xmath30 , @xmath18 , @xmath31 , and @xmath19 bands .
an analysis of hst data by mighell & burke ( 1999 ) reveals the same picture : the stars of ursa minor are about 14 gyr old , formed in a single burst that lasted less than 2 gyr , and are metal - poor with a metallicity @xmath32 } \approx -2.2 $ ] dex .
all of these characteristics indicate that the stars in ursa minor formed in a single burst during the earliest stages of the formation of our galaxy .
ursa minor has little or no gas : searches for neutral hydrogen by young ( 2000 ) and for ionized hydrogen by gallagher ( 2003 ) yield no positive detections , only placing upper limits of @xmath33 on hi and @xmath34 on hii .
the absence of gas is consistent with the old stellar population in ursa minor ; however , the question of why ursa minor quickly stopped forming stars whereas the similar dsph carina did not remains unanswered . with the possible exception of sagittarius ,
ursa minor has the greatest flattening among the known galactic dsphs .
ih derive an ellipticity of @xmath35 and a position angle for the major axis of @xmath36 degrees .
kleyna ( 1998 ) derive similar values .
the ellipticity of ursa minor is distinctly larger than those of the other dsphs ( again excluding sagittarius ) , which range from 0.1 to 0.35 ( ih ) .
the left - hand panel in figure [ fields ] shows a @xmath37 section of the sky in the direction of ursa minor from the digitized sky survey .
the dashed and solid ellipses delineate the core and tidal radii of ursa minor from ih .
hartwick & sargent ( 1978 ) and aaronson ( 1983 ) were the first to measure radial velocities for 12 stars in ursa minor , finding a systemic velocity of about @xmath38 km s@xmath1 .
hargreaves ( 1994 ) find a mean heliocentric radial velocity of @xmath39 km s@xmath1 from the radial velocities of 35 giants .
armandroff , olszewski , & pryor ( 1995 ) combined the data from hargreaves ( 1994 ) and from olszewski , aaronson , & hill ( 1995 ) with their own measurements to obtain mean radial velocities for a sample of 94 stars .
excluding one possible non - member , the mean heliocentric radial velocity is @xmath40 km s@xmath1 .
our study adopts this last value .
the adopted value is in a reasonable agreement with the most recent determination of this quantity by wilkinson ( 2004 ) , who derive @xmath41 km s@xmath1 from the measured radial velocities for 162 stars .
the dispersion of velocities within a dsph provides information about its mass and how the mass is distributed .
aaronson & olszewski ( 1987 ) report a velocity dispersion for ursa minor of @xmath42 km s@xmath1 from radial velocities of seven stars .
this dispersion indicated for the first time that the mass greatly exceeds that expected from stars alone .
subsequent measurements have confirmed the large dispersion : hargreaves ( 1994 ) find @xmath43 km s@xmath1 from a sample of 35 giants and armandroff , olszewski , & pryor ( 1995 ) find @xmath44 km s@xmath1 from a sample of 93 giants .
the larger sample of radial velocities in wilkinson ( 2004 ) provides information on the dependence of velocity dispersion on the projected radius .
the bottom panel of their figure 1 shows that the velocity dispersion is constant at about 12 km s@xmath1 to a radius of approximately 32 arcmin ; beyond this radius , the velocity dispersion drops sharply to about 2 km s@xmath1 .
hargreaves ( 1994 ) and armandroff , olszewski , & pryor ( 1995 ) assume virial equilibrium , that light follows mass , and the luminosity and structural parameters from ih to calculate a mass - to - light ratio in solar units , @xmath45 , of @xmath46 and @xmath47 , respectively .
the uncertainty for the second of these values includes only the contribution from the velocity dispersion , whereas the uncertainty for the first value includes the contributions from all of the relevant parameters . making the same assumptions and using one of the above velocity dispersions , ih , kleyna ( 1998 ) , and carrera ( 2002 )
infer similarly large values of @xmath48 : @xmath49 , @xmath50 , and 70 , respectively .
in contrast , palma ( 2003 ) find @xmath51 .
they derive a surface density profile from a sample of stars selected to be members of ursa minor using three - color washington - band photometry .
fitting a king model to this profile produces a tidal radius of @xmath52 arcmin . this value is larger than those derived by ih and kleyna ( 1998 ) , which affects the derived mass and luminosity of the galaxy .
palma ( 2003 ) also note that the @xmath53 could be even smaller if the velocity dispersion tensor of the dsph is anisotropic .
richstone & tremaine ( 1986 ) caution that the @xmath53 calculated using king s method will be overestimated or underestimated depending on whether the velocity dispersion is larger in the radial or tangential directions , respectively . despite a lack of any evidence for an anisotropic dispersion tensor , palma ( 2003 ) conclude that the @xmath48 of ursa minor may be as low as 16 .
the larger distance , hence larger luminosity , found by bellazzini ( 2002 ) further reduces the lower limit on the @xmath48 to 7 .
palma ( 2003 ) note that a power law with an index of @xmath54 is a better fit to the surface density profile of ursa minor at radii between 20 arcmin and 100 arcmin than is a king model .
they argue that such a profile indicates that the galactic tidal force has produced a halo of escaping stars around the dsph .
they also argue that a @xmath53 near the lower limit allowed by the uncertainty in the mass estimate is more consistent with a picture in which `` this system is very likely undergoing significant mass loss due to its tidal interaction with the milky way . ''
gmez - flechoso & martnez - delgado ( 2003 ) have made the above argument more quantitative .
they derive @xmath55 for ursa minor by evolving an @xmath56-body model of the dsph in a potential of the milky way on an orbit constrained by the proper motion of the dsph from schweitzer , cudworth , & majewski ( 1997 ) , varying the @xmath53 of the dsph in order to match the surface brightness profile of the model dsph with the observed profile .
the study finds that an acceptable match exists only if ursa minor has @xmath48 within a factor of two of @xmath57 .
gmez - flechoso & martnez - delgado ( 2003 ) argue that their lower value for the @xmath53 is consistent with the values derived from the velocity dispersion if the maximum possible effect from radial anisotropy is taken into account .
they also argue that the observed velocity dispersion `` could be inflated by the effects of the substructures in the main body '' or by departures from virial equilibrium due to the galactic tidal force .
the numerical simulations of piatek & pryor ( 1995 ) show that , if the galactic tidal force causes a departure from virial equilibrium , then escaping stars produce a velocity gradient an apparent rotation along the projected major axis of the dsph .
hargreaves ( 1994 ) and armandroff , olszewski , & pryor ( 1995 ) both found a statistically significant velocity gradient , but along the minor axis instead of the major axis . subtracting the velocity gradient from the radial velocities
does not reduce the measured velocity dispersion significantly .
kroupa ( 1997 ) and klessen & kroupa ( 1998 ) show that the velocity gradient is hidden if the tidal debris is aligned along the line of sight to the dsph .
palma ( 2003 ) rule out this alignment because it would produce more broadening of the horizontal branch than they observe .
thus , a departure from virial equilibrium can not explain the difference between the @xmath53 found by gmez - flechoso & martnez - delgado ( 2003 ) and those found using the measured velocity dispersion .
olszewski & aaronson ( 1985 ) discovered a statistically significant change in surface density across their 3 arcmin @xmath27 5 arcmin field and interpreted this variation as evidence for substructure in ursa minor .
it is now known that the highest peak in the surface density of the dsph is near the eastern edge of this field .
demers ( 1995 ) also detect a statistically significant clump of stars at the same location , which causes a step in the surface density profile at a radius of about 1 arcmin .
the deep _ hst _ photometry of battinelli & demers ( 1999 ) supports the existence of this clump and may also suggest the presence of additional structure within the clump .
other studies of substructure have focused on a possible secondary peak or `` shoulder '' in the surface density profile , first noted by ih , that is about 15 arcmin northeast of the primary peak .
kleyna ( 1998 ) , eskridge & schweitzer ( 2001 ) , and palma ( 2003 ) demonstrate that a `` shoulder '' in the profile is statistically significant , though a secondary peak is not .
kleyna ( 2003 ) used measured radial velocities for stars in ursa minor to demonstrate the presence of a distinct sub - population with a velocity dispersion of 0.5 km s@xmath1 much smaller than the 8.8 km s@xmath1 of the whole sample of velocities at the location of the `` shoulder . ''
excluding the stars in this substructure from the sample would increase the measured velocity dispersion of the dsph and , thus , the inferred @xmath53 .
thus , the presence of substructure does not reconcile the @xmath53 found by gmez - flechoso & martnez - delgado ( 2003 ) with that derived from the velocity dispersion .
two pictures of ursa minor emerge from the above discussion .
( 1 ) dark matter is the main component of the mass of ursa minor and therefore it determines the structure and internal dynamics of the dsph .
( 2 ) ursa minor contains little or no dark matter and therefore the galactic tidal force has had and continues to have an important effect on the structure and internal dynamics of the dsph . knowing the proper motion of ursa minor can help to discriminate between these two pictures by constraining the orbit and , thus , the strength of the galactic tidal force on ursa minor .
if the perigalacticon is small , ursa minor would need to contain a large amount of dark matter to have survived until now .
the presence of tidal signatures need not indicate that the dsph is out of virial equilibrium or that it contains only luminous matter .
conversely , a large perigalacticon would require a low @xmath53 for the dsph so that the galactic tidal force could have produced the observed signatures .
a proper motion for ursa minor also tests the hypothesis that it is a member of a `` stream '' of objects in the galactic halo that share a similar orbit .
lynden - bell & lynden - bell ( 1995 ) propose that ursa minor , the large magellanic cloud ( lmc ) , the small magellanic cloud ( smc ) , draco , and possibly also sculptor and carina , form a stream .
they propose that such a stream forms from the fragments of a larger , tidally disrupted galaxy .
the theory of streams in the galactic halo is falsifiable because it predicts a proper motion for each member of the stream .
there is some controversy regarding the reality of this stream : anguita , loyola , & pedreros ( 2000 ) measured the proper motion of the lmc and found that it is inconsistent with the prediction of lynden - bell & lynden - bell ( 1995 ) , whereas the proper motions measured by jones , klemola , & lin ( 1994 ) and kroupa & bastian ( 1997 ) are consistent .
there are two , independent , published results for the proper motion of ursa minor .
scholz & irwin ( 1993 ) used palomar sky survey plates for the first and second epochs , which are separated by about 26 years , and plates taken with the tautenburg schmidt telescope for the third epoch which provide a total time baseline of about 31 years .
they find a proper motion , `` as is on the sky , '' of @xmath58 mas cent@xmath1 ( the last row in their table 3 , but using an uncertainty 80 instead of 20 mas cent@xmath1 for @xmath59 , as implied by the text ) .
schweitzer ( 1996 ) and schweitzer , cudworth , & majewski ( 1997 ) used a total of 39 photographic plates taken with the palomar 5-meter hale telescope ( first epoch ) and the kpno 4-meter telescope ( second and third epochs ) .
the time between the first- and third - epoch images is about 42 years .
schweitzer ( 1996 ) reports a proper motion of @xmath60 mas cent@xmath1 .
the proper motions measured by these two studies agree within their uncertainties . here
we present a third independent measurement of the proper motion of ursa minor .
the measurement derives from images of two distinct fields , each containing a known quasi - stellar object ( qso ) , taken with _ hst_. sections [ data ] , [ analysis ] , and [ pm ] describe the data , the derivation of mean centroids at each epoch , and the derivation of the proper motion from the centroids , respectively .
the last of these sections contains a comparison of our measured proper motion with those determined by scholz & irwin ( 1993 ) and schweitzer ( 1996 , see also schweitzer , cudworth , & majewski 1997 ) .
section [ orbitsec ] derives orbital elements in a realistic potential for the milky way .
section [ sec : disc ] discusses the implications of the orbit of ursa minor for : 1 ) its membership in proposed structures in the galactic halo ; 2 ) its star formation history ; 3 ) the mass of the milky way ; and 4 ) the importance of the galactic tidal force and , thus , whether the dsph contains dark matter .
finally , section [ sec : summary ] summarizes our main results .
the data comprise images of two distinct fields in the direction of ursa minor taken at three epochs with _ hst_. a known qso is at , or close to , the center of each field .
the left panel in figure [ fields ] shows the locations of the two fields on the sky .
both fields are within the core radius and close to the minor axis . _ hst _ imaged the larger of the two fields , umi j@xmath61 , on february 16 , 2000 ; february 15 , 2001 ; and february 1 , 2002 using the space telescope imaging spectrograph ( stis ) with no filter ( 50ccd ) .
ursa minor was in the continuous viewing zone during the first two epochs and was not during the last .
each of the eight dither positions has six images at the first two epochs , for a total of 48 images per epoch , while each dither position has only three , for a total of 24 images , at the last epoch .
the average exposure times for the three epochs are 176 s , 176 s , and 210 s , respectively .
the qso in this field is at @xmath62 ( j2000.0 ) . from bellazzini ( 2002 ) ,
its @xmath18- and @xmath19-band magnitudes are 20.3 and 19.7 , respectively .
figure [ fig : qso - stis ] shows a spectrum of the qso , which indicates a redshift of 1.216 .
the top - right panel in figure [ fields ] shows a sample image of this field from the 2000 epoch . the image is the average of six images at one dither position with cosmic rays removed .
the cross - hair indicates the location of the qso .
_ hst _ imaged the smaller of the two fields , umi j@xmath63 , on march 14 , 1999 ; march 10 , 2001 ; and march 2 , 2003 using the planetary camera ( pc ) of the wide field and planetary camera 2 ( wfpc2 ) and the f606w filter .
there are 40 images for the first epoch , 36 for the second , and 36 for the third .
the exposure time is 160 s for each image . here , the qso is at @xmath64 ( j2000.0 ) , its @xmath18- and @xmath19-band magnitudes are @xmath65 and @xmath66 , respectively ( bellazzini 2002 ) , and figure [ fig : qso - wfpc ] shows its spectrum .
the spectrum implies a redshift of 0.716 .
the bottom - right panel of figure [ fields]depicts an average of three images in this field from the 1999 epoch , with cosmic rays removed .
the cross - hair indicates the location of the qso .
note the scarcity of stars in both sample images .
bristow ( 2004 ) notes that the decreasing charge transfer efficiency in the stis and wfpc2 ccds may induce a spurious contribution to the proper motion measured with the method described by piatek ( 2002 : p02 hereafter ) and piatek ( 2003 ; p03 hereafter ) .
the cosmic particle radiation damages the crystal lattice of a detector , creating `` charge traps , '' and the number of these traps increases with time . during the readout , charge moves along the @xmath67 axis : `` up , '' or towards the increasing @xmath67-values , for stis , and `` down , '' or in the direction of decreasing @xmath67-values , for pc2 .
charge moves `` left , '' towards decreasing @xmath68-values in the serial register .
when a `` packet '' of charge corresponding to some object encounters a trap , it loses some of its charge .
a passing packet partially fills a trap , so a subsequent packet loses less charge .
streaks of light along the @xmath68 and @xmath67 axes seen trailing behind objects in the images are the visual artefacts caused by the gradual release of charge from the traps .
the loss of charge for motion in the @xmath68 direction is smaller than for that in the @xmath67 direction ; it is negligible for stis ( brown 2002 ) and one third as large for wfpc2 ( heyer 2004 ) .
the subsequent discussion focuses only on the effect in the @xmath67 direction .
the loss of charge to traps causes both the flux and the psf of an object to depend on its @xmath67 coordinate in the image .
the farther a packet travels , the more charge it loses .
hence , the measured flux of an object far from the serial register is smaller than that for an identical object closer to the serial register ; in other words , the measured flux becomes position dependent .
this effect is immaterial to our method .
however , as the packets representing an object move along the @xmath67 axis , those on its leading side fill partially each trap encountered , so that there are fewer traps available to remove charge from subsequent packets .
this non - uniform loss of charge across the object changes its psf .
consequently , the centroid of the packet shifts in the direction opposite to the readout direction .
the shift is greatest for objects that are farthest from the serial register .
it is also larger for faint objects than for bright objects ( bristow & alexov 2002 ) .
this effect changes the measured proper motion only because the number of charge traps increases with time , thus causing the shift of a centroid to also change with time .
the analysis of bristow ( 2004 ) warns that this spurious proper motion may be comparable to the actual proper motion of a dsph .
there are two possible approaches to removing the effect of the spurious shifts of the centroids along the @xmath67 axis with time . in the first approach ,
the images are corrected by restoring the trapped charge to its original packet .
this correction requires that the distribution and the properties of the charge traps be known .
bristow & alexov ( 2002 ) constructed a physical model of the charge traps and then wrote a computer code that makes the aforementioned correction for an image taken with stis . unless stated otherwise , the results presented are for corrected images .
no comparable software exists for wfpc2 , unfortunately . a second approach to correcting for the decreasing charge transfer efficiency
is including a term that varies linearly with @xmath67 in the equations which transform the coordinates of an object at different epochs to a common coordinate system .
the coefficients of these transformations are fitted as described in section [ sec : dpm ] .
the analysis of the data taken with wfpc2 uses this approach .
the analysis of the data and derivation of the proper motion of ursa minor are basically the same as those described in p02 .
p03 provides some additional insight into the method .
this article only lists the principal steps here and comments where it is appropriate on several changes in the details of the analysis .
the principal steps of the analysis are : ( 1 ) using the daophot and allstar software packages ( stetson 1987 , 1992 , 1994 ) , determine an initial estimate of the centroid of each object stars and the qso . ( 2 ) construct an effective point - spread function ( epsf ; anderson & king 2000 ) for each epoch separately from a select set of stars and the qso .
( 3 ) iteratively , fit the epsf to an image of an object using least - squares to derive a more accurate centroid of the object and , using these more accurate centroids , derive a more accurate epsf . repeat until a stable solution ensues .
there may be a maximum of @xmath56 centroids for an object , where @xmath56 is the number of images .
( 4 ) transform the centroids measured at one epoch to a common fiducial coordinate system and calculate the average .
the fiducial coordinate system for an epoch is the @xmath69 coordinate system of the chronologically first image .
( 5 ) determine the transformation between the fiducial coordinate systems of the later epochs and that of the first epoch using stars which are likely members of ursa minor and common to all three epochs .
the fit for the transformation simultaneously determines the change with time of the coordinates of some objects , @xmath70 and @xmath71 in pixel yr@xmath1 .
ideally , these objects are not members of ursa minor .
the set of objects whose motion is fit is built starting with the qso and iteratively adding one object at a time , in order of descending @xmath72 for the scatter about the mean coordinate , until the highest @xmath72 of an object not yet in the set is below a specified limit .
the shifts are with respect to stars of ursa minor which , by definition , have @xmath73 .
this procedure for deriving the values of @xmath70 and @xmath71 for objects is new to our method ; section [ pm ] below discusses its details .
the advantage of this new method over that used by p02 and p03 is that it more accurately includes the contribution from the uncertainty in the coordinate transformation between epochs to the uncertainty in the proper motion .
( 6 ) derive the proper motion of ursa minor from the @xmath70 and @xmath71 of the qso . before calculating the transformation to the fiducial coordinate system ,
the centroids measured in each image taken with wfpc2 are corrected for the 34th - row defect ( shaklan , sharman , & pravdo 1995 ; anderson & king 1999 ) .
in addition , all centroids are corrected for the known optical distortions in the wfpc2 and stis instruments .
this paper uses the most recent corrections for wfpc2 derived by anderson & king ( 2003 ) .
the corrections for stis continue to be those in the stis data handbook ( brown 2002 ) .
the following subsections present and discuss the key diagnostics of the performance of our method .
equation 22 in p02 defines a flux residual diagnostic , @xmath74 , which is a quantitative measure of how the shape of an object matches the epsf . ignoring random noise , for a perfect match , @xmath75 ; otherwise , @xmath76 , where the sign depends on the details of the mismatch and the size of the mismatch increases with the brightness of the star . a plot of @xmath77 as a function of location can help unravel a dependence of true psf with location in an image .
our method uses a single , location - independent epsf for deriving centroids of objects in all images for a given field and epoch .
if the true psf of an object varies from image to image , then the @xmath77s for this object will show a scatter , for example around @xmath78 if the epsf represents some average of the true psfs .
the plot of @xmath77 as a function of location will show trends if the true psf varies with location in the image .
figure [ rf - stis ] shows plots of @xmath77 as a function of location for the umi j@xmath61 field .
the left - hand panels show plots of @xmath77 _ versus _ @xmath68 and the right - hand panels show plots of @xmath77 _ versus _ @xmath67 . from top to bottom , the rows of plots are for epochs 2000 , 2001 , and 2002 .
each plot combines points from all images for a given epoch .
the solid square marks the points corresponding to the qso . in all six panels ,
the values of @xmath77 for the qso are negative and distinct from those for the other objects stars .
the panel [ stis-10-rf - x ] and , to a lesser extent , panel [ stis-9-rf - x ] show a linear trend between @xmath74 and @xmath68 .
@xmath77 tends to be positive around @xmath79 pixel and negative around @xmath80 pixel .
no other panels show a trend of @xmath77 with @xmath68 or @xmath67 .
figure [ rf - stis ] implies the following .
( 1 ) the psf of the qso has a different shape than that of a star .
the @xmath77s for the qso become comparable to those for the stars when the value of the central pixel in the @xmath81 science data array for the qso ( see p02 for a description of the structure of the data ) is arbitrarily reduced by about 20% .
thus , the psf of the qso is narrower than that of the stars , perhaps because the qso is bluer than the stars .
( 2 ) the psf at a given location in the field varies from image to image , _
i.e. _ , with time .
this variation causes the scatter in the @xmath77s of a given object .
( 3 ) the psf varies across the field for epoch 2002 and , to a lesser degree , for epoch 2001 because the plots for those epochs show trends with location .
figure [ rf - pc ] is analogous to figure [ rf - stis ] for the umi j@xmath63 field . from the top row to the bottom , the plots correspond to epochs 1999 , 2001 , and 2003 , respectively .
all of the plots show that the @xmath77s for the qso are negative , with most of the points outside of the plot for epoch 2001 .
the mean @xmath77 of the qso at any epoch is more negative than the mean for any other object .
for example , the object at @xmath82 is a star as bright as the qso but its @xmath77s tend to be less negative or even positive .
again , the likely reason for the distinctness of the qso is its narrower psf .
no plot shows an evident trend of @xmath77 with @xmath68 or @xmath67 ; however , the small number of objects contributing to these plots makes this conclusion uncertain .
figures [ stis - rxry ] and [ pc - rxry ] plot the position residual , @xmath83 or @xmath84 , of a centroid as a function of its location within a pixel pixel phase , @xmath85 or @xmath86 . the position residuals are @xmath87 and @xmath88 , respectively , where @xmath89 and @xmath90 are the components of the mean centroid in the fiducial coordinate system the system of the first image in time at a given epoch .
the pixel phases are @xmath91 and @xmath92 , where the function @xmath93 returns the integer part of a variable @xmath94 . in the presence of only random noise ,
the points in figures [ stis - rxry ] and [ pc - rxry ] would scatter symmetrically around zero and would not exhibit any trends .
panels [ stis-8-rxry ] , [ stis-9-rxry ] , and [ stis-10-rxry ] in figure [ stis - rxry ] plot , from top to bottom , @xmath83 _ versus _ @xmath85 , @xmath84 _ versus _
@xmath86 , @xmath83 _ versus _
@xmath86 , and @xmath84 _ versus _ @xmath85 .
the panels are for the umi j@xmath61 field , epochs 2000 , 2001 , and 2002 , respectively .
no plot in any of the panels in figure [ stis - rxry ] shows a trend of position residual with pixel phase .
figure [ pc - rxry ] is analogous to figure [ stis - rxry ] for the umi j@xmath63 field .
panels [ pc-7-rxry ] , [ pc-9-rxry ] , and [ pc-10-rxry ] correspond to epochs 1999 , 2001 , and 2003 , respectively .
no plot in figure [ pc - rxry ] shows a trend of position residual with pixel phase .
the figures show that the values of @xmath83 and @xmath84 for the qso are indistinguishable from those for the stars
. the smaller scatter of the values for the qso reflects its status as one of the brightest objects in the field .
the similar distributions of the points for the qso and the stars implies that the narrower psf of the qso , compared to that for the stars , does not affect the accuracy of its measured centroid .
kuijken & rich ( 2002 ) show that the precision of a centroid determined by psf fitting is proportional to @xmath95 times the fwhm of the psf .
the constant of proportionality is approximately 0.67 .
if there is no source of error other than the uncertainty in the intensity registered by the pixel , then a plot for the entire sample of the @xmath96 scatter of the measured centroids for an object around their mean as a function of the @xmath95 of the object should consist of points falling on a straight line passing through the origin . if the points fall above this straight line at large @xmath97 , they indicate the presence of additive uncertainty , either random or systematic , that is independent of the signal .
the distribution of points above and below the line is affected by the sampling uncertainty in the @xmath96 and variations in the fwhm of the psf .
the psf can vary with location in the field , from image to image thus with time , or both .
figure [ stis-8-rms - sn ] is a plot of the @xmath96 of the @xmath68-component of a centroid ( top panel ) and of the @xmath67-component ( bottom panel ) as a function of @xmath95 for the epoch 2000 umi j@xmath61 field .
figures [ stis-9-rms - sn ] and [ stis-10-rms - sn ] are the same for epochs 2001 and 2002 , respectively . note that the figures have different horizontal and vertical scales .
the solid line in each plot is the best - fitting function of the form @xmath98 where @xmath99 is the @xmath96 scatter in either the @xmath68 or @xmath67 directions and @xmath100 and @xmath101 are free parameters .
table 1 tabulates the fitted values of @xmath100 and @xmath101 .
each point has equal weight in the fit . in all of the plots ,
the adopted functional form is a good fit to the points and the best fit requires a non - zero @xmath101 .
the fitted slopes are in approximate agreement with the value expected for the 1.5 pixel fwhm of our epsf .
the points corresponding to the qso are not farther from the fitted line than those corresponding to bright stars , which argues that the difference between the psf of the qso and of a star does not affect the @xmath96 scatter of the measured centroid of the qso .
figures [ pc-7-rms - sn ] , [ pc-9-rms - sn ] , and [ pc-10-rms - sn ] are the corresponding plots of @xmath96 _ versus _
@xmath95 for the umi j@xmath63 field epochs 1999 , 2001 , and 2003 , respectively .
note that the figures have different horizontal and vertical scales .
the solid lines show that equation [ rms - sn ] is again a good fit to the points .
table 1 contains the best - fitting @xmath100 and @xmath101 for all three epochs . the figures and best - fitting values are similar to those for the other field . the additive uncertainty , @xmath101 , inferred from the fits shown in figures [ stis - rms - sn ] and [ pc - rms - sn ] significantly increases the uncertainty of the final average centroid for objects with high @xmath97 .
p02 note that the @xmath96 scatter of measured centroids about their mean is smaller for measurements at a single dither position than for those at multiple dither positions .
the interpretation of this was that the errors producing the additive uncertainty depend primarily on pixel phase , which argues that they arise from errors in the shape of the epsf .
thus , the measurements of the centroid at a single dither position are not independent when the @xmath97 is high .
both p02 and p03 calculate the larger uncertainty in the mean centroid resulting from the smaller number of independent measurements by assuming that , when the @xmath96 of a component of the centroid approaches the corresponding @xmath102 , the uncertainty in the average centroid approaches the @xmath96 divided by the square - root of the number of dither positions instead of the square - root of the number of measurements ( see equation 29 in p02 ) .
additional testing while analyzing data for this article shows that the method of p02 and p03 overestimates the uncertainties in the centroids for those objects with high @xmath97 , while underestimating them for those with low @xmath97 .
the usual estimator for the standard deviation about the mean is biased downwards when the sample size is small ( sclove 2004 ) .
for example , the bias is a factor of 0.89 for a sample size of three .
this partly explains why p02 finds a smaller @xmath96 scatter around the mean for measured centroids at a given dither position compared to the scatter for the centroids at all of the dither positions together . a better method to estimate the uncertainty in the final average centroid than that used by p02 , including the effects of a lack of independence of measurements at a given dither position , is to use the mean centroid , @xmath103 , at each dither position @xmath104 . the uncertainty in either the @xmath94 or @xmath105 component of the final centroid is @xmath106 where @xmath107 is the number of dither positions and @xmath108 is the mean of the mean centroids at the dither positions
the uncertainty estimated with equation [ eq : uncnew ] is typically 20%
40% larger than the usual estimate from all of the measured centroids for an object with a @xmath97 larger than about 15 .
this increase is less than the approximately 70% increase implied by the procedure of p02 , but the increase extends to a lower @xmath97 .
this article uses equation [ eq : uncnew ] to calculate the uncertainty in the mean centroid of an object at one epoch .
the procedure for deriving a proper motion in this article is different from that described in p02 and p03 .
section [ sec : dpm ] below describes this new procedure in detail for the three - epoch data taken either with stis or wfpc2 .
the subsequent sections describe the actual realization of this procedure for the case of ursa minor .
the three elements in determining the relative motion of the qso with respect to the stars of the dsph are : 1 ) a transformation between the fiducial coordinate systems of the later epochs and that of the first epoch the `` standard coordinate system '' determined by objects common to all three epochs .
the transformation contains a translation , @xmath109 , a rotation , @xmath110 , and a ratio between the two scales , @xmath111 . here
@xmath112 is the index denoting the epoch and , for a transformation , may be 2 or 3 .
2 ) a mean position for an object @xmath113 , @xmath114 , which is the average of the three measured centroids in the standard coordinate system . the transformed standard coordinates at epochs 2 and 3 are related to the mean coordinate by @xmath115 for the qso and for any object with a large @xmath116 for the scatter around @xmath114 with @xmath117 . here
@xmath118 is the uniform linear motion of the object in the standard coordinate system in pixel yr@xmath1 and @xmath119 is the time of the epoch @xmath112 measured from @xmath120 .
3 ) a simultaneous fit for the coefficients of the transformations , the @xmath114 , and the @xmath118 using all of the measured centroids of objects common to the three epochs . simultaneously fitting for the @xmath118 ensures that the stars of the dsph remain at rest in the standard coordinate system .
let @xmath121 be the measured coordinates and their uncertainties of the centroid of object @xmath113 in the fiducial coordinate system of epoch @xmath112 , where @xmath122 , 2 , or 3 .
the transformation of these measured coordinates to the standard coordinate system is @xmath123 the offset @xmath124 defines the pivot point for the rotation : it is @xmath125 pixel for stis and @xmath126 pixel for wfpc2 . the @xmath67
coordinate in equations [ eq : tranx ] and [ eq : trany ] , @xmath127 , includes a correction for the shift caused by the charge traps in the ccd . as discussed at the end of section [ data ] , the equations @xmath128 approximately correct for the shift in the @xmath105 direction induced by the charge traps .
equation [ eq : cti - stis ] is for the data taken with stis , whereas equation [ eq : cti - pc ] is for those taken with wfpc2 .
these equations are necessary only when the method of bristow & alexov ( 2002 ) has not been used to restore the images .
because charge traps affect a faint object more than a bright object , an object contributes to the determination of the fitted parameter @xmath129 , and has its coordinates subsequently corrected , only if its @xmath97 is smaller than some specified limit .
for the three - epoch data , equations [ eq : tranx ] [ eq : tranuy ] and [ eq : cti - stis ] or [ eq : cti - pc ] contain nine fitted parameters .
their values result from minimizing a @xmath72 of the form @xmath130.\ ] ] the @xmath114 in the above equations add an additional @xmath131 fitted parameters , where @xmath56 is the number of objects in the fit .
however , the @xmath114 can be calculated analytically .
the minimization procedure starts by fitting for the @xmath118 of the qso while assigning @xmath132 pixel yr@xmath1 for all of the other objects .
the procedure then iteratively selects the object with the largest contribution to the @xmath72 and fits for its @xmath118 together with all of the previously fit parameters .
the iteration terminates when the highest @xmath72 among the objects not yet selected is smaller than some specified limit .
the proper motion of a dsph derives from @xmath133 for the qso .
the uncertainty in @xmath134 for the qso and , thus , the uncertainty in the proper motion of the dsph comes from increasing the @xmath72 by one above the minimum ( _ e.g. _ , press 1992 , ch .
one component of the motion changes away from its fitted value , with all other parameters adjusted to minimize @xmath72 , until the total @xmath72 increases by one .
the difference between this value and the nominal value is the uncertainty for this component of the motion .
the procedure repeats for the other component of the uncertainty .
a correct determination of the uncertainty in @xmath134 using this method requires that the uncertainties in the @xmath135 be realistic .
a break - down of this assumption would be indicated by a minimum total @xmath116 that is significantly larger than one per degree of freedom .
stis was the imaging detector in the umi j@xmath61 field at all of the epochs .
thus , it was possible to correct the images for the effects induced by the charge traps in the ccd using the method of bristow & alexov ( 2002 ) . correcting the images
eliminates the need to fit for the free parameter @xmath129 in equation [ eq : cti - stis ] .
the number of objects with measured centroids is 81 for the first epoch , 49 for the second , and 64 for the third . among these , 32 are common to the three epochs .
the choice for the individual @xmath116 that triggers fitting for uniform linear motion is 15 .
there are approximately 4 degrees of freedom per object when the motion is not fit , so this limit should be triggered by chance for only 0.15 object in a sample of 32 . the final fitted transformation and motion of the qso
are not sensitive to the exact value of this limit .
the total @xmath116 from equation [ eq : k2 ] is much larger than one per degree of freedom when using the uncertainties estimated with equation [ eq : uncnew ] .
thus , there must be an uncertainty that arises from an error that is the same for all measurements made at a single epoch , but changes significantly between epochs .
figure [ fig : k2stis ] explores whether this uncertainty is additive or multiplicative by plotting the individual contribution of each object to the @xmath116 _ versus _ @xmath97 . from top to bottom , the panels show the contribution to @xmath116 for measurements at the first epoch only , second only , third only , and at all of the epochs .
the plots show that @xmath116 is , on average , the same at all values of @xmath97 . increasing the uncertainties given by equation
[ eq : k2 ] with an additional additive uncertainty would decrease the @xmath116 values only at high @xmath97 , so figure [ fig : k2stis ] indicates that the uncertainty given by equation [ eq : uncnew ] should be multiplied by a constant instead of having a constant added in quadrature .
multiplying the uncertainties by 1.55 produces a @xmath116 of one per degree of freedom .
this choice for the multiplicative factor ensures that the uncertainties in the proper motions calculated as described at the end of section [ sec : dpm ] reflect the true scatter of the points about the fitted transformation and proper motions .
the @xmath116 values in figure [ fig : k2stis ] are calculated with the increased uncertainty .
we can point to no likely source for the additional error operating between epochs .
figure [ fig : stis - rxry ] plots position residuals , defined for an object @xmath113 by @xmath136 as a function of location in the standard coordinate system . here
the subscript @xmath137 indicates that the residual is for a centroid from the @xmath138 epoch transformed to the standard coordinates system of the time of the first epoch . from top to bottom , the panels are for the first , second , and third epoch , respectively .
the panels in the left column show @xmath139 _ versus _ @xmath68 and those in the right column show @xmath140 _ versus _ @xmath67 . in all of the plots ,
the points scatter around the horizontal axis ; no plot shows a trend between @xmath139 and @xmath68 or @xmath140 and @xmath67 or a systematic bias towards either positive or negative residuals .
the scatter is a few hundredths of a pixel . although not shown in the figure , the plots of @xmath139 _ versus _ @xmath67 and of @xmath140 _ versus _
@xmath68 do not show any trends either .
figure [ xqyqstis ] shows the location of the qso as a function of time in the standard coordinate system .
the top panel shows the variation of the @xmath68 coordinate and the bottom panel does the same for the @xmath67 coordinate .
the motion of the qso is @xmath141 pixel yr@xmath1 .
the contribution to the total @xmath116 from the qso is 0.17 , which is reflected in the close agreement between the points and the straight lines in figure [ xqyqstis ] .
the contribution to the @xmath116 has approximately two degrees of freedom , which implies an 8% probability of a @xmath116 smaller than 0.17 by chance .
wfpc2 was the imaging detector for this field .
because no corresponding software exists that restores an image taken with this detector , equation [ eq : cti - pc ] must be used to account for the shifts in the @xmath67 direction caused by the charge traps .
the number of objects with measured centroids is 21 for the 1999 epoch , 20 for 2001 , and 19 for 2003 . among these , 19 are common to all of the epochs .
this is a small number , which proved to be insufficient to precisely determine the value of @xmath129 in the fitting procedure the value was large and had the wrong sign .
however , @xmath129 depends on the number of charge traps as a function of time rather than on the observed field .
thus , @xmath129 is derived from images of a field in fornax , which contain almost 200 objects and were taken within a few days of those for umi j@xmath63 .
the result is @xmath142 , with the correction applying to objects with a @xmath143 .
equation [ eq : cti - pc ] for umi j@xmath63 has @xmath129 held constant at this value and applied to objects with @xmath143 .
the choice for the individual @xmath116 that triggers fitting for uniform motion is 30 .
this results in fitting a motion for only the qso and one star with a large motion .
we chose not to fit a motion for any other stars because of the small sample size , though the final fitted transformation and motion of the qso are not sensitive to fitting motions for a few additional objects .
the @xmath116 per degree of freedom from equation [ eq : k2 ] is again larger than one when using the uncertainties from equation [ eq : uncnew ] .
figure [ fig : k2pc ] plots the contribution to @xmath116 _ versus _
@xmath97 for objects in umi j@xmath63 . as in figure
[ fig : k2stis ] , the lack of a trend implies that the additional error is multiplicative .
a multiplier of 1.413 yields a @xmath116 per degree of freedom of one .
figure [ fig : pc - rxry ] plots the position residuals @xmath139 and @xmath140 for the umi j@xmath63 field , similar to figure [ fig : stis - rxry ] .
no panel shows a trend of @xmath139 with @xmath68 or @xmath140 with @xmath67 . although not depicted in the figure , there are no trends between @xmath139 and @xmath67 or @xmath140 and @xmath68 either .
figure [ xqyqpc ] shows the location of the qso as a function of time in the standard coordinate system for the umi j@xmath144 field .
note that the slopes of the corresponding plots in figures [ xqyqpc ] and [ xqyqstis ] need not be the same because the two fields are rotated with respect to each other .
the motion of the qso is @xmath145 pixel yr@xmath1 .
the contribution to the total @xmath116 from the qso is 0.01 , which is reflected in the close agreement between the points and the straight lines in figure [ xqyqpc ] .
the contribution to the @xmath116 has approximately two degrees of freedom , which implies a 0.5% probability of a @xmath116 smaller than 0.01 by chance .
the small contributions to the @xmath116 by the qso for both the stis and wfpc2 data suggest that our estimated uncertainties in the centroids are too small
. however , other objects in our fields with fitted motions do not have unusually small @xmath116 values see tables 3a and 3b presented in section [ measuredpm ] . a smaller threshold for the contribution to @xmath116 which triggers fitting for a motion would reduce the multiplier needed to make the total @xmath116 per degree of freedom equal to one , which would increase the contribution to the @xmath116 from the qso .
we choose not to reduce our thresholds further because the small number of stars in both of our fields could produce a systematic error in the measured centroids depending on position .
the small number both forces us to use an epsf that is constant in both space and time within an epoch and limits our ability to detect systematic errors .
it also limits the amount of information available to determine the transformation to the standard coordinate system .
we prefer to average over possible systematic errors rather than spuriously remove them by fitting motions .
we then think that it is best to increase the uncertainties in the measured centroids so that the uncertainty in the fitted motion of the qso reflects the typical scatter of the centroids around the fitted transformation and motions even if the centroids of the qso have a smaller scatter .
we again emphasize that the measured motion of the qso does not depend strongly on the sample of objects with fitted motions .
table 2 records the measured proper motion for each field in the equatorial coordinate system and their weighted mean .
the uncertainty in the proper motion of the dsph depends on both the uncertainty in the measured motion of the qso in the standard coordinate system and on the length of the time baseline .
the baseline for the umi j@xmath61 field is about 2 yrs , whereas that for the umi j@xmath144 field is about 4 yrs . because of the small number of objects in the latter field , we chose to give the two fields more nearly equal weight in the average .
we did this by doubling the uncertainty for the proper motion derived from the umi j@xmath144 field and this is the uncertainty listed in table 2 .
the proper motion in table 2 is that measured by a heliocentric observer and , thus , includes the effects of the motions of the lsr and of the sun with respect to the lsr .
the measured proper motion is the best quantity to use for comparisons with the other independent measurements .
there are two previous measurements of the proper motion for ursa minor .
scholz & irwin ( 1993 ) report a measured proper motion of @xmath146 mas cent@xmath1 ( the last row in their table 3 , but using the larger uncertainty for @xmath59 implied in the text ) and schweitzer ( 1996 ; see also schweitzer , cudworth , & majewski 1997 ) reports a value of @xmath147 mas cent@xmath1 .
the four rectangles in figure [ mus - comp ] depict the four independent measurements of the proper motion .
the center of a rectangle is the best estimate of the proper motion and the sides are offset from the center by the 1-@xmath99 uncertainties .
rectangles 1 and 2 represent the measurements by scholz & irwin ( 1993 ) and schweitzer ( 1996 ) , respectively , whereas rectangles 3 and 4 represent the measurements from this study for the fields umi j@xmath61 and umi j@xmath63 , respectively . our measurements , 3 and 4 , agree within their uncertainties .
the @xmath148 components of measurements 2 and 3 disagree and at least one of these two measurements must be affected by systematic errors larger than the quoted uncertainties . our other measurement , 4 , is closer to 3 than to 2 . because of its large uncertainty
, measurement 1 does not provide much additional information on the proper motion .
the weighted mean of our measurements 3 and 4 , listed in the bottom line of table 2 , differs from measurement 2 by 3.0@xmath27 the uncertainty in the difference of @xmath148 and 0.8@xmath27 the uncertainty in the difference of @xmath59 .
thus , these two reported measurements do not agree within their uncertainties and one or the other or both must have systematic errors larger than the quoted uncertainties . through experimentation , we have eliminated the following sources for the disagreement : 1 ) the exact shape of the epsf . using an analytic epsf and one derived from archival data sets for a field in the globular cluster @xmath149 centauri had little effect on the final proper motion .
2 ) the value for parameter @xmath129 in equation [ eq : cti - pc ] , which approximately corrects for the effect of the increasing number of charge traps in the pc2 ccd .
large and arbitrary changes in the value can not simultaneously reconcile measurements 2 , 3 , and 4 .
3 ) changes in the number of stars which determine the transformation between epochs .
the fitted parameters , including the motion of the qso , do not change significantly if objects with apparently discrepant measurements _ i.e. _ , having a large @xmath116 after being fit for a uniform motion are excluded from the sample .
4 ) the exact value of the limit on @xmath116 which triggers fitting for a uniform motion .
thus , we are unable to explain the origin of the difference between our measured proper motion and that of schweitzer ( 1996 ; see also schweitzer , cudworth , & majewski 1997 ) .
table 3a tabulates the proper motions for those objects in the umi j@xmath61 field for which it was measured .
table 3b does the same for the umi j@xmath63 field .
the first line of each table corresponds to the qso and subsequent objects are listed in order of decreasing @xmath97 .
the i d number of an object is in column 1 , the @xmath68 and @xmath67 coordinates of an object in the earliest image of the first epoch ( o5bl01010 for umi j@xmath61 and u50j0101r for umi j@xmath63 ) are in columns 2 and 3 , and the @xmath97 of the object at the first epoch is in column 4 .
the components of the measured proper motion , expressed in the equatorial coordinate system , are in columns 5 and 6 .
each value is the measured proper motion in the standard coordinate system corrected by adding the weighted mean proper motion of ursa minor given in the bottom line of table 2 . to indicate that this correction has been made ,
the proper motion of the qso is given as zero .
the listed uncertainty of each proper motion is the uncertainty of the measured proper motion , calculated in the same way as for the qso , added in quadrature to that of the average proper motion of the dsph .
the contribution of the object to the total @xmath116 is in column 7 .
the proper motion is unreliable if this value is much larger than 2.0 .
removing the contributions to the measured proper motion from the motion of the lsr and the peculiar motion of the sun , yields the galactic rest frame proper motion .
this proper motion would be measured by a hypothetical observer at the location of the sun and at rest with respect to the galactic center .
columns ( 2 ) and ( 3 ) in table 4 list the components of the galactic rest frame proper motion expressed in the equatorial coordinate system , @xmath150 , for our two fields .
the derivation of @xmath150 assumes : 220 km s@xmath1 for the circular velocity of the lsr ; 8.5 kpc for the distance of the sun from the galactic center ; and @xmath151 km s@xmath1 ( dehnen & binney 1998 ) for the peculiar velocity of the sun , where the components are positive if @xmath152 points radially away from the galactic center , @xmath153 is in the direction of rotation of the galactic disk , and @xmath154 points in the direction of the north galactic pole . for convenience , columns ( 4 ) and ( 5 ) in table 4 list the components of the galactic rest frame proper motion in the galactic coordinate system , @xmath155 .
the next three columns list the components of the space velocity in the cylindrical coordinate system centered on the dsph .
the derivation of these assumes a heliocentric distance of 76 kpc to ursa minor .
the components of the space velocity are positive if @xmath156 points radially away from the galactic rotation axis ; @xmath157 is in the direction of rotation of the galactic disk ; and @xmath158 points in the direction of the north galactic pole .
the final two columns list the radial and tangential components of the space velocity with respect to a hypothetical observer at rest at the galactic center . column ( 9 ) gives the radial component , which is positive if it points away from the galactic center and column ( 10 ) gives the tangential component .
knowing the space velocity of a dsph permits a determination of its orbit for a given form of the galactic potential .
this study adopts a galactic potential that has a contribution from a disk of the form ( miyamoto & nagai 1975 ) @xmath159 from a spheroid of the form ( hernquist 1990 ) @xmath160 and from a halo of the form @xmath161 in the above equations , @xmath162 is the galactocentric distance , @xmath31 is the projection of @xmath162 onto the plane of the galactic disk , and @xmath158 is the distance from the plane of the disk .
all other quantities in the equations are adjustable parameters and their values are the same as those adopted by johnston , sigurdsson , & hernquist ( 1999 ) : @xmath163 m@xmath164 , @xmath165 m@xmath164 , @xmath166 km s@xmath1 , @xmath167 kpc , @xmath168 kpc , @xmath169 kpc , and @xmath170 kpc .
figure [ orbit ] shows the projections of the orbit of ursa minor resulting from an integration of the motion in the galactic potential given by equations [ diskpot ] , [ spherpot ] , and [ logpot ] .
the integration extends for 3 gyr backwards in time and begins at the current location of ursa minor with the negative of the space velocity given in the bottom line of table 4 .
the solid square marks the current location of the dsph , the solid star indicates the center of the galaxy , and the small open circles mark the points where @xmath171 or , in other words , where the orbit crosses the plane of the galactic disk .
the large open circle is for reference : it has a radius of 30 kpc . in the right - handed coordinate system of figure [ orbit ] ,
the current location of the sun is on the positive @xmath68-axis .
the figure shows that ursa minor is close to apogalacticon and has both a moderately inclined and eccentric orbit .
table 5 enumerates the elements of the orbit of ursa minor .
the value of the quantity is in column ( 4 ) and its @xmath7 confidence interval is in column ( 5 ) .
the latter comes from 1000 monte carlo experiments , where an experiment integrates the orbit using an initial velocity that is chosen randomly from a gaussian distribution whose mean and standard deviation are the best estimate of the space velocity and its quoted uncertainty , respectively .
the eccentricity of the orbit , defined as @xmath172 is 0.39 , though the 95% confidence interval ranges from a nearly circular orbit to a nearly radial orbit with a perigalacticon near 10 kpc .
the inclination , @xmath173 , implies a retrograde orbit .
the inclination of the nominal orbit is about 56 degrees to the galactic plane , though a nearly polar orbit is within the 95% confidence interval .
the longitude of the ascending node , @xmath174 , is measured counter - clockwise from the positive @xmath68-axis .
as outlined in section [ intro ] , ursa minor may be a member of a stream that also includes the lmc , smc , draco , and , possibly , sculptor and carina . if ursa minor is a member , then its predicted measured proper motion is @xmath175 mas cent@xmath1 , or @xmath176 mas cent@xmath1 with a position angle of 328 degrees ( lynden - bell & lynden - bell 1995 ) .
the measured proper motion from this study is @xmath177 mas cent@xmath1 with a position angle of @xmath178 degrees , which are @xmath179 and @xmath180 away from the predicted values , respectively .
we rule out the possibility that ursa minor is a member of the proposed stream at more than @xmath181 .
kroupa , theis , & boily ( 2004 ) show that the 11 dwarf galaxies nearest to the milky way are nearly on a plane , whose pole is at @xmath182 degrees . adopting the direction of the angular momentum vector as the pole of the orbit ,
then the location of the pole is @xmath183 because of the left - handed nature of the galactic rotation , prograde orbits have @xmath184 and retrograde orbits have @xmath185 .
thus , the pole of our orbit for ursa minor is @xmath186 degrees , where the uncertainties are 1-@xmath99 values from the monte carlo simulations .
thus , the motion of ursa minor is not in the plane defined by the nearby dwarf galaxies . studies of the stellar population in ursa minor , described in section [ intro ] , indicate that the dsph contains low - metallicity , old stars . the majority of stars formed close to a hubble time ago , suggesting that ursa minor lost its gas quickly .
in contrast , carina had extensive star formation about 7 gyr ago , which continued to within 1 gyr ago .
these two galaxies have nearly the same luminosity and surface brightness and , thus , presumably should have retained gas to a similar degree . since they have not , is it because their galactic orbits are different ?
the answer appears to be no .
table 4 of p03 shows that carina has orbital elements similar to those of ursa minor .
the perigalacticon of carina is probably smaller than that of ursa minor ( a nominal value of 20 kpc _ versus _ 40 kpc ) , so carina should have lost its gas more rapidly to stronger tidal shocks ( mayer 2001 ) and larger ram pressure stripping ( blitz & robishaw 2002 ; gallart 2001 ; mayer & wadsley 2003 ) . a notable difference between the two orbits is that that for carina is prograde , while that for ursa minor is retrograde .
however , how this difference could have affected star formation is unclear .
ursa minor is bound gravitationally to the milky way .
the galactocentric space velocity of the dsph imposes a lower limit on the mass of the milky way within the present galactocentric radius of the dsph , @xmath31 . assuming a spherically symmetric mass distribution and zero for the total energy of the dsph , the lower limit for the mass of the milky way is given by @xmath187 setting @xmath188 kpc and using the values from table 4 for @xmath189 and @xmath190 , @xmath191 .
this lower limit is consistent with other recent estimates of the mass of the milky way , such as the mass of @xmath192 within @xmath193 kpc found by sakamoto , chiba , & beers ( 2003 ) .
the milky way potential adopted in section [ orbitsec ] has a mass of @xmath194 out to @xmath188 kpc .
the measured ellipticity of ursa minor is one of the largest known for the dsphs . if the galactic tidal force deformed ursa minor from an initial spherical shape to its present elongated shape , then the position angle of its projected major axis should be similar to the position angle of the galactic - rest - frame proper motion vector
the position angle of the projected major axis is @xmath195 degrees and the position angle of the galactic - rest - frame proper motion vector is @xmath196 degrees .
the difference between the two position angles is 2.6 times its uncertainty , arguing that ursa minor is not elongated along its orbit .
gmez - flechoso & martnez - delgado ( 2003 ) derive an @xmath48 for ursa minor in the range 6 to 24 by matching the radial profile of a model dsph from n - body simulations to the observed radial profile . in the simulations , the dsph moves on the orbit found by schweitzer , cudworth , & majewski ( 1997 ) .
given our different orbit for ursa minor , is this range of @xmath48 values still compatible with the observed limiting radius of the radial profile ?
a poor - man s substitute for numerical simulations is to calculate the tidal radius , @xmath197 , beyond which a star becomes unbound from the dsph . for a logarithmic galactic potential , @xmath197
is given by ( king 1962 ; oh , lin , & aarseth 1992 ) @xmath198\ln[(1+e)/(1-e ) ] + 1 } \ , \frac{m}{m_g}\right)^{1/3 } a. \label{eq : rtidal}\ ] ] here @xmath199 is the eccentricity of the orbit , @xmath100 is the semi - major axis ( @xmath200 ) , @xmath201 is the mass of the dsph , and @xmath202 is the mass of the galaxy within @xmath100 . equating @xmath197 with the observed limiting radius derived by fitting a king ( 1966 ) model , @xmath203 , yields a value for @xmath48 for a given orbit .
if @xmath204 arcmin , then 50% of the orbits in monte carlo simulations have @xmath205 .
this @xmath48 is just within the range quoted by gmez - flechoso & martnez - delgado ( 2003 ) .
if @xmath206 arcmin , then 50% of the orbits have @xmath207 , and 95% have @xmath205 .
thus , the @xmath48 derived by gmez - flechoso & martnez - delgado ( 2003 ) is incompatible with the larger measured value of @xmath206 arcmin for our measured proper motion for ursa minor . on the other hand ,
any of the larger values for @xmath48 derived from the observed velocity dispersion ( see section [ intro ] ) is more compatible with the larger value of @xmath203 .
equation [ eq : rtidal ] shows that @xmath208 , so the values for @xmath53 derived using this equation are sensitive to the measured value of the limiting radius and the identification of that radius with the tidal radius . until kinematic measurements definitively identify the tidal radius , an @xmath53 derived with the above argument
should be treated with caution .
the average measured @xmath48 for galactic globular clusters is 2.3 ( pryor & meylan 1993 ) .
could the @xmath48 of ursa minor be this low ?
numerical simulations by oh , lin , & aarseth ( 1995 ) and piatek & pryor ( 1995 ) show that the ratio of the limiting radius derived by fitting a theoretical king model ( king 1966 ) , @xmath203 , to the tidal radius defined by equation ( [ eq : rtidal ] ) is a useful indicator of the importance of the galactic tidal force on the structure of a dsph .
these simulations show that : if @xmath209 , the galactic tidal force has little effect on the structure of the dsph ; at @xmath210 , the effect of the force increases rapidly with increasing @xmath211 ; and , for @xmath212 , the dsph disintegrates in a few orbits . assuming that @xmath213 and @xmath204 arcmin , @xmath214 for 59% of the orbits generated in monte carlo simulations . if @xmath206 arcmin , the fraction is 99.6% .
thus , it is unlikely that ursa minor would have survived a hubble time on its current orbit if it did not contain any dark matter .
\1 ) two independent measurements of the proper motion for ursa minor produce a weighted - average value of @xmath215 mas century@xmath1 in the equatorial coordinate system for a heliocentric observer .
our value and the proper motion of schweitzer ( 1996 ; also schweitzer , cudworth , & majewski 1997 ) disagree by more than twice the uncertainty in the difference .
\2 ) removing the contributions of the motion of the sun and of the lsr to the measured proper motion , gives a galactic - rest - frame proper motion of @xmath216 mas century@xmath1 in the equatorial coordinate system for an observer at the location of the sun but at rest with respect to the galactic center . in the galactic coordinate system , this motion is @xmath217 mas century@xmath1 .
\4 ) the best estimate for the orbit shows that ursa minor is close to its apogalacticon of @xmath220 kpc and is moving closer to the milky way on a retrograde orbit with eccentricity @xmath221 and a smallest angle between the orbital plane and the plane of the galactic disk of @xmath222 .
the closest approach to the galaxy , perigalacticon , is @xmath223 kpc .
ursa minor completes one full circuit around the milky way in @xmath224 gyr .
\5 ) ursa minor is not a likely member of the `` stream '' of galaxies on similar orbits proposed by lynden - bell & lynden - bell ( 1995 ) , nor is its orbit confined to the plane of satellite galaxies noted by kroupa , theis , & boily ( 2004 ) .
we thank daniel eisenstein and jim liebert for obtaining new spectra for the qsos in ursa minor after the discovery spectra were lost .
we thank the referee , dr . van altena , for helpful suggestions .
cp and sp acknowledge the financial support of the space telescope science institute through the grants hst - go-07341.03-a and hst - go-08286.03-a .
ewo acknowledges support from the space telescope science institute through the grants hst - go-07341.01-a and hst - go-08286.01-a and from the national science foundation through the grants ast-9619524 and ast-0098518 .
mm acknowledges support from the space telescope science institute through the grants hst - go-07341.02-a and hst - go-08286.02-a and from the national science foundation through the grant ast-0098661 .
dm is supported by fondap center for astrophysics 15010003 .
aaronson , m. 1983 , apj , 266 , l11 aaronson , m. , & olszewski , e. w. 1987 , in dark matter in the universe , iau symposium no . 117 , edited by j. kormendy and g. r. knapp ( reidel , dordrecht ) , p. 153
anderson , j. , & king , i. r. 1999 , pasp , 111 , 1095 anderson , j. , & king , i. r. 2000 , pasp , 112 , 1360 anderson , j. , & king , i. r. 2003 , pasp , 115 , 113 anguita , c. , loyola , p. , & pedreros , m. h. 2000 , aj , 120 , 845 armandroff , t. e. , olszewski , e. w. , & pryor , c. 1995 , aj , 110 , 2131 battinelli , p. , & demers , s. 1999 , aj , 117 , 1764 bellazzini , m. , ferraro , f. r. , origlia , l. , pancino , e. , monaco , l. , & oliva , e. 2002 , aj , 3222 bennett , c. l. 2003 , apjs , 148 , 1 blitz , l. , & robishaw , t. 2000 , apj , 541 , 675 bristow , p. 2004 , ce - stis-2004 - 003 bristow , p. , & alexov , a. 2002 , ce - stis - isr 2002 - 001 brown , t. , , 2002 , in @xmath225 stis data handbook version 4.0 , ed .
b. mobasher ( baltimore : stsci ) carrera , r. , aparicio , a. , martnez - delgado , d. , & alonso - garca , j. 2002 , aj , 123 , 3199 cudworth , k. m. , olszewski , e. w. , & schommer , r. a. 1986 , aj , 92 , 766 dehnen , w. , & binney , j. j. 1998 , mnras , 298 , 387 demers , s. , battinelli , p. , irwin , m. j. , & kunkel , w. e. 1995 , mnras , 274 , 491 eskridge , p. b. , & schweitzer , a. e. 2001 , aj , 122 , 3106 gallagher , j. s. , madsen , g. j. , reynolds , r. j. , grebel , e. k. , & smecker - hane , t. a. 2003 , apj , 588 , 326 gallart , c. , martnez - delgado , d. , gmez - flechoso , m. a. , & mateo , m. 2001 , aj , 121 , 2572 gmez - flechoso , m. a. , & martnez - delgado , d. 2003 , apj , 586 , l123 gratton , r. g. , fusi pecci , f. , carretta , e. , clementini , g. , corsi , c. e. , & lattanzi , m. 1997 , apj , 491 , 749 hargreaves , j. c. , gilmore , g. , irwin , m. j. , & carter d. 1994 , mnras , 271 , 693 hartwick , f. d. a. , & sargent , w. l. w. 1978 , apj , 221 , 512 hernquist , l. 1990 , apj , 356 , 359 heyer , biretta , et al .
2004 , wfpc2 instrument handbook , version 9.0 ( baltimore : stsci ) .
irwin , m. , & hatzidimitriou , d. 1995 , mnras , 277 , 1354 ( ih ) johnston , k. v. , sigurdsson , s. , & hernquist , l. 1999 , mnras , 302 , 771 jones , b. f. , klemola , a. r. , & lin , d. n. g. 1994 , aj , 107 , 1333 king , i. r. 1962 , aj , 67 , 471 king , i. r. 1966 , aj , 71 , 64 klessen , r. s. , & kroupa , p. 1998
, apj , 498 , 143 kleyna , j. t. , geller , m. j. , kenyon , s. j. , kurtz , m. j. , & thorstensen , j. r. 1998 , aj , 2359 kleyna , j. t. , wilkinson , m. i. , gilmore , g. , & wyn evans , n. 2003 , apj , l21 kroupa , p. 1997 , newa , 2 , 139 kroupa , p. , & bastian , u. 1997 , newa , 2 , 77 kroupa , p. , theis , c. , & boily , c. m. 2004 , a&a , in press kuijken , k. , & rich , r. m. 2002 , aj , 124 , 2054 lynden - bell , d. , & lynden - bell , r. m. 1995 , mnras , 275 , 429 martnez - delgado , d. , alonso - garca , j. , aparicio , a. , gmez - flechoso , m. a. 2001 , apj , 549 , l63 mateo , m. 1998 , ara&a , 36 , 435 mayer , l. , governato , f. , colpi , m. , moore , b. , quinn , t. , wadsley , j. , stadel , j. , & lake , g. 2001 , apj , 559 , 754 mayer , l. , & wadsley 2003 , astro - ph/0309073 mighell , k. j. , & burke , c. j. 1999 , aj , 118 , 366 milgrom , m. 1983 , apj , 270 , 365 miyamoto , m. , & nagai r. 1975 , pasj , 27 , 533 nemec , j. m. , wehlau , a. , & mendes de oliveira , c. 1988 , aj , 96 , 528 oh , k. s. , lin , d. n. c. , & aarseth , s. j. 1992 , apj , 386 , 506 oh , k. s. , lin , d. n. c. , & aarseth , s. j. 1995 , apj , 442 , 1420 olszewski , e. w. , & aaronson , m. 1985 , aj , 90 , 2221 olszewski , e. w. , aaronson , m. , & hill , j. m. 1995 , aj , 110 , 2120 palma , c. , majewski , s. r. , siegel , m. h. , patterson , r. , ostheimer , j. , & link , r. 2003 , aj , 1352 piatek , s. , & pryor , c. 1995 , aj , 109 , 1071 piatek , s. , pryor , c. , olszewski , e. w. , harris , h. c. , mateo , m. , minniti , d. , monet , d. g. , morrison , h. , & tinney , c. g. 2002 , aj , 124 , 3198 ( p02 ) piatek , s. , pryor , c. , olszewski , e. w. , harris , h. c. , mateo , m. , minniti , d. , & tinney , c. g. 2003 , aj , 126 , 2346 ( po3 ) press , w. h. , teukolsky , s. a. , vetterling , w. t. , & flannery , b. p. 1992
, numerical recipes , the art of scientific computing , second edition , ( cambridge : cambridge university press ) pryor , c. , & meylan , g. 1993 , in structure and dynamics of globular clusters , edited by s. djorgovski & g. meylan ( asp , san francisco ) , p. 357
reid , i. n. 1997 , aj , 114 , 161 richstone , d. o. , & tremaine , s. 1986 , aj , 92 , 72 sakamoto , t. , chiba , m. , & beers , t. c. 2003 , a&a , 397 , 899 scholz , r. d. , & irwin , m. j. 1993 , in iau symp .
161 , astronomy from wide - field imaging , ed . h. t. macgillivray ( dordrecht : kluwer ) , p. 535
schweitzer , a. e. 1996 , ph .
d. thesis , univ .
wisconsin schweitzer , a. e. , cudworth , k. m. , & majewski , s. r. 1997 , in asp conf .
127 , proper motions and galactic astronomy , ed .
r. m. humphreys ( san francisco : asp ) , 132 sclove , s. l. 2004 , lecture notes ( www.uic.edu/classes/idsc/ids571/samplvar.pdf ) shaklan , s. , sharman , m. c. , & pravdo , s. h. 1995 , appl .
, 34 , 6672 stetson , p. b. 1987 , pasp , 99 , 191 stetson , p. b. 1992 , in asp conf .
25 , astronomical data analysis software and systems , ed .
d. m. worrall , c. biemesderfer , & j. barnes ( san francisco : asp ) , 297 stetson , p. b. 1994 , pasp , 106 , 250 schweitzer , a. e. , cudworth , k. m. , & majewski , s. r. 1997 , in asp conf .
127 , proper motions and galactic astronomy , ed .
r. m. humphreys ( san francisco : asp ) , 132 van den bergh , s. 2000 , the galaxies of the local group ( cambridge : cambridge university press ) wilkinson , m. i. , kleyna , j. t. , wyn evans , n. , gilmore , g. f. , irwin , m. j. , & grebel , e. k. 2004 , apj , 611 , l21 wilson , a. g. 1955 , pasp , 67 , 27 young , l. m. 2000 , aj , 119 , 188 lccccc & 2000 & 0.879 & 0.023 & 0.874 & 0.017 + umi@j@xmath61 & 2001 & 0.768 & 0.040 & 0.846 & 0.027 + & 2002 & 1.004 & 0.015 & 0.882 & 0.020 + & 1999 & 0.921 & 0.024 & 0.881 & 0.020 + umi@j@xmath63 & 2001 & 0.974 & 0.021 & 0.926 & 0.024 + & 2003 & 0.987 & 0.025 & 0.878 & 0.030 + ccccccc 1 & 514 & 513 & 207 & @xmath232 & @xmath233 & 0.17 + 2 & 671 & 640 & 47 & @xmath234 & @xmath235 & 2.58 + 3 & 321 & 375 & 44 & @xmath236 & @xmath237 & 1.86 + 4 & 771 & 685 & 38 & @xmath238 & @xmath239 & 4.72 + 5 & 497 & 831 & 34 & @xmath240 & @xmath241 & 0.66 + 6 & 318 & 665 & 23 & @xmath242 & @xmath243 & 0.25 + 7 & 689 & 623 & 11 & @xmath244 & @xmath245 & 0.96 + 8 & 453 & 509 & 10 & @xmath246 & @xmath247 & 1.27 + lrrrrrrrrr umi@j@xmath61&@xmath252&@xmath253&@xmath254&@xmath255&@xmath256&@xmath257&@xmath258&@xmath259&@xmath260 + umi@j@xmath63&@xmath261&@xmath262&@xmath263&@xmath264&@xmath265&@xmath266&@xmath267&@xmath268&@xmath269 + weighted mean:&@xmath270&@xmath271&@xmath272&@xmath273&@xmath274&@xmath275&@xmath276&@xmath277&@xmath278 + lcccc perigalacticon&@xmath279&kpc&@xmath280&@xmath5 + apogalacticon&@xmath281&kpc&@xmath282&@xmath6 + eccentricity & @xmath199&&@xmath283&@xmath284 + period&@xmath285&gyr&@xmath286&@xmath287 + inclination&@xmath173°&124&@xmath288 + longitude&@xmath174° & 153&@xmath289 | this article presents a measurement of the proper motion of the ursa minor dwarf spheroidal galaxy determined from images taken with the hubble space telescope in two distinct fields .
each field contains a quasi - stellar object which serves as the `` reference point . ''
the measured proper motion for ursa minor , expressed in the equatorial coordinate system , is @xmath0 mas century@xmath1 .
removing the contributions of the solar motion and the motion of the local standard of rest yields the proper motion in the galactic rest frame : @xmath2 mas century@xmath1 .
the implied space velocity with respect to the galactic center has a radial component of @xmath3 km s@xmath1 and a tangential component of @xmath4 km s@xmath1 . integrating
the motion of ursa minor in a realistic potential for the milky way produces orbital elements .
the perigalacticon and apogalacticon are 40 @xmath5 kpc and 89 @xmath6 kpc , respectively , where the values in the parentheses represent the @xmath7 confidence interval derived from monte carlo experiments .
the eccentricity of the orbit is @xmath8 and the orbital period is @xmath9 gyr .
the orbit is retrograde and inclined by @xmath10 degrees to the galactic plane .
ursa minor is not a likely member of a proposed stream of galaxies on similar orbits around the milky way . nor is the plane of its orbit coincident with a recently - proposed planar alignment of galaxies around the milky way . comparing the orbits of ursa minor and
carina shows no reason for the different star formation histories of these two galaxies .
ursa minor must contain dark matter to have a high probability of surviving disruption by the galactic tidal force until the present . |
for centuries people combined different solid objects to construct other objects . on this way
the restriction of a solid body being impenetrable by another body produces various packing problems .
one example is packing spheres with a famous application to the liquid of hard spheres with the topological disorder still not understood .
intuitively , solid infinite cylinders should be more cumbersome because they can be entangled . indeed , unlike compact bodies such as spheres
, the infinite cylinders can touch one another at any distance , producing infinitely long spatial correlations .
this may explain why for the whole history the first attempt to deal with the cylinders ( finite rigid rods ) was made by lars onsager only in 1949 .
it has long been known how to make mutually touch seven half - infinite cylinders , yet , the configuration essentially makes use of the possibility of three half - infinite cylinders to lie in one plane while touching both each other and the pivotal cylinder with their stumps .
placing other three cylinders right over them produces the configuration needed .
however , being half - infinite is crucial for topology : a half - infinite cylinder does not `` puncture '' the space . in 1986 , in connection with the continuum percolation problem , we analyzed the possibility of arranging mutually touching arbitrary infinite cylinders invoking the method of degrees of freedom @xcite .
the method prompted only possible solutions to be found either numerically or analytically .
we predicted that seven mutually touching cylinders might exist and the configuration might not be rigid . in 2004
we discovered @xcite that the configuration of seven mutually touching straight round infinite cylinders indeed had a numerical solution ( while solving 10 equations for 10 variables ) , which we named 7-knot .
the configuration keeps its topology when one remaining degree of freedom ( predicted in @xcite being the scissor angle @xmath0 ( see fig .
[ fig:1 ] ) between two chosen cylinders : the pivotal 0th cylinder ( red ) and the 1st cylinder ( brown ) in fig .
[ fig:2 ] ) , changes .
the rigorous nonlinear equations for the angles and positions ( see below ) , which provide orientations of each cylinder to be touching all the others were solved with a renown mathcad program with a computational accuracy of order of @xmath1 ( all relevant files with the texts of the mathcad programs as well as animations are given in online resources 1 - 8 ) ) .
the method was extended to regular lattices of mutually touching cylinders to produce an artificial matter with the ultimate poisson ratio -1 because it could either shrink or expand as a whole due to cylinder entanglement @xcite . .
@xmath2 axis that extends to the right is not shown .
( a ) the 1st cylinder touching the pivotal vertical cylinder of unit radius @xmath3
. letters @xmath4 and @xmath5 mark the directions along @xmath6 .
( b ) two cylinders touching each other and the pivotal 0th cylinder , scaledwidth=85.0% ] ) . @xmath7
marks the pivotal cylinder and @xmath8 marks the first cylinder as given in fig .
[ fig:1 ] . here
the scissor angle was @xmath9 .
the configuration has @xmath10 unknotted cylinders each of which can be removed without disturbing other cylinders , scaledwidth=85.0% ] in @xcite the 7-knot was first found in a configuration topology that precluded having all the cylinders to be equal in radii ( congruent ) . in 2009
we succeeded @xcitein discovering other topologies and published the result with the 7-knot of the mutually touching infinite cylinders of equal radii ( the congruent one , here equipped with asterisk : 7 * -knot ) ( see fig .
[ fig:1]b ) in @xcite and fig .
[ fig:3 ] ) .
however , probably because the results of constructing pure mathematical bodies were embedded into the physical context , the findings remained unknown for the mathematical community .
a recent paper @xcite rediscovered our finding of mutually touching seven infinite equal radii cylinders , the authors being unaware of our earlier results with cylinders of equal and non - equal radii and of the nuances of their topology .
it may illustrate a still existing gap between the worlds of physics and mathematics though both are intimately connected through the plato s world @xcite .
meantime , the obvious difference between 7-knot and 7 * -knot pushed us forward in curiosity questions about the topology of entanglement of the cylinders that constitute 7-knots , the limits of angles and radii where they exist , calculating the latitudinal and longitudinal angles while the scissor angle @xmath0 sweeps from the zero value to @xmath11 etc .
also there is the question of quantification of the difference in the topology between the configurations of the 7-knot with a set of non - equal radii that can be continuously reduced to 7 * -knot with equal cylinders , and the basins of sets of non - equal radii that can not be continuously reduced to the equal radii .
the switching between the topologies always happens through infinity while the cylinders try to become parallel which is a kind of degeneracy .
therefore some orientational entanglement takes place which might be important to be understood on the basic level of straight infinite cylinders . here , we reveal the detailed structure of 7-knot along with our finding that natural extension to 8-and 9-knots does exist , as predicted in @xcite .
we computed the 8-knot and the ultimate 9-knot .
these novel configurations of mutually touching round infinite cylinders lack the angular degrees of freedom because of having too many mutual contacts .
it is not at all obvious that such rigid configurations should exist and simple comparison with the circles and spheres below illustrate the whole problem of the degrees of freedom for 7- , 8- , and 9- knots .
finally , we mention some possible consequences of cylinder entanglement for the growing field of various string - like objects in physics and possible connection with rodger penrose s
twistors @xcite .
let us begin with circles to explain the method of controlling configuration by the number @xmath12 of degrees of freedom . for n circles on the plane being in mutual contacts the number
@xmath12 satisfies the degree of freedom equation ( dfe ) : @xmath13 indeed , each circle has @xmath14 degrees of freedom ( @xmath15,@xmath16 positions of its center on the plane ) , that is @xmath17 in total for all the freedom and @xmath18 stands for pairwise restrictions due to mutual contacts .
the last term is the number of solid degrees of any two dimensional configuration as a whole : it can be rotated @xmath19 and translated @xmath20 .
easy to see that for @xmath21 eq .
( [ eq:1 ] ) gives @xmath22 .
this rigid configuration actually exists for any radii : one can always mutually contact @xmath23 circles .
if we restrict ourselves with outside contacts then we would have a single configuration with a mirror counterpartner . anyway
mirror degrees do not count here because they are discrete . then let us take @xmath24 which gives @xmath25 .
having @xmath25 indicates that the fourth circle should have a predetermined radius ; others are changeable ( so - called apollonius circles ) .
the next case is more dramatic : @xmath26 gives us @xmath27 .
this should be enough because we still have @xmath28 radii to be adjusted as we please ( more precisely , four radii , because one radius should be a measuring unit ) .
however , such an ultimate ( one can not put @xmath29 because @xmath30 is out of range , only @xmath28 radii out of @xmath31 can be changeable ) configuration does not exist .
it means that even if a configuration is not forbidden by the counting of the degrees of freedom , it is not at all obvious that it can exist . ) ; ( b ) the same as in ( a ) with calculated parameters ( the tolerance to zero was set @xmath32 ) .
the configuration has @xmath23 unknotted cylinders each of which can be removed to infinity without disturbing other cylinders .
therefore the configuration is topologically different from the configuration in fig .
[ fig:2 ] . the animation file with the sweeping angle @xmath0
is included in online resource 1 as well as the animation of the solid rotation of the configuration in online resource 6,scaledwidth=85.0% ] for spheres instead of eq .
( [ eq:1 ] ) one has @xmath33 for @xmath24,@xmath22 , which is also rigid at some changeable radii . for n=5 eq .
( [ eq:2 ] ) gives @xmath25 , which is quite realizable provided that one sphere radius is determined by the rest four .
however , @xmath29 gives @xmath27 that does not exist though we still have enough radii to fix .
the ultimate limit would be @xmath34 with @xmath30 , which is still enough though also does not exist .
let us see what occurs for infinite cylinders .
here the situation is different : all configurations predicted by dfe exist .
we posed this question with the connection to the topology of the percolation cluster in @xcite where the corresponding dfe was discussed ( see also @xcite with more relevant examples ) .
the equation for n cylinders is @xcite , @xcite : @xmath35 eq .
( [ eq:3 ] ) looks like a minor alternation as compared with eq .
( [ eq:2 ] ) but it takes into account the crucially important affine degree of freedom , that for infinite cylinders implies that a cylinder axis can always punch a two dimensional plane at a point ( @xmath14 degrees for @xmath15,@xmath16 ) and can be ray - like oriented ( @xmath14 degrees for two angles @xmath0 and @xmath36 , see fig . [ fig:1 ] ) . eq .
( [ eq:3 ] ) always predicts actually existing configurations .
one of them was 7-knot @xcite , @xcite,@xcite and the other two will be shown below to be 8-knot and 9-knot .
let us use eq .
( [ eq:3 ] ) for @xmath34 . from eq .
( [ eq:3 ] ) we have @xmath37 which means that 7-knot is not rigid : one degree ( it can be the scissor angle @xmath0 ) remains .
the calculations indeed proved this @xcite , @xcite,@xcite .
for @xmath38 we obtain @xmath39 so two radii of the cylinders should be determined and they are , as we show below .
then , @xmath40 gives @xmath30 with only @xmath23 cylinders of changeable radii ( again only two in reality as far as one radius always goes for a measuring unit ) .
this ultimate configuration exists , in spite of the non - existence of ultimate configurations for circles and spheres .
we write down the equations for the mutual contacts in the vector notations .
the number of equations we choose to solve is @xmath41 ( see below ) which makes @xmath42 , @xmath43 , and @xmath44 in case of 7- , 8- , and 9-knot , respectively .
we solve them numerically with the mathcad solver ( all relevant texts of the programs given in online resources 2 - 5 ) .
the sets of equations also could be simplified more as to their number and variables .
for example , in @xcite after some analytical work we used only @xmath45 equations to fix topology and to solve only for angles but in @xcite we used either @xmath45 or @xmath42 equations with the same result ( see below ) .
our practice shows that the solution converges nearly as fast even for 9-knot with @xmath44 equations , or for 8-knot with @xmath43 equations , as for 7-knot with either @xmath42 or @xmath45 equations depending on how one defines the variables .
the basic equations for the contacting cylinders can be obtained from fig .
[ fig:1 ] .
we use the 0th cylinder of the radius @xmath3 , taken to be a measuring unit , as the pivotal one which axis lies along @xmath46 .
all other cylinders are constructed as touching it by default .
the 1st cylinder of radius @xmath47 touching the 0th cylinder is described by the unit vector @xmath48 that defines the orientation of the cylinder and the vector @xmath49 connecting the origin with a point on the cylinder axis .
the longitudinal angle @xmath36 is counted counterclockwise from the direction of @xmath6 axis , marked by letter @xmath4 , to the direction of the vertical projection of the cylinder axis onto the plane @xmath50 , marked by letter @xmath51 in fig.[fig:1]a .
the axis @xmath2 , not shown in fig .
[ fig:1 ] , extends to the right .
the latitudinal angle @xmath0 is shown in fig . [ fig:1]a
. then we define @xmath52 @xmath53 in such a representation it is easy to check from eqs .
( [ eq:4 ] ) , ( [ eq:5 ] ) that the vector @xmath54 which length is @xmath55 ( fig .
[ fig:1]a ) is orthogonal to both @xmath46 axis and @xmath48 , therefore the 1st cylinder is indeed touching the 0th cylinder by construction .
any point on the axis of the 1st cylinder is described by the vector @xmath56 , where @xmath57 is the affine parameter along the cylinder axis .
thus , as far as we fixed the pivotal cylinder to be vertical , all other cylinders are now characterized by only 3 degrees of freedom each ( provided we keep the radii fixed ) : ( @xmath0,@xmath36,@xmath58 ) , ( @xmath59,@xmath60,@xmath61 ) , etc . to fix the configuration
completely , we always keep @xmath62 and @xmath63 , which removes 2 degrees of freedom .
now we have @xmath64 cylinders to make touch mutually .
that allows one to rewrite dfe of eq .
( [ eq:3 ] ) in the form : @xmath65 where @xmath12 is of course identical to the one from eq .
( [ eq:3 ] ) . yet instead of @xmath43 contact equations for @xmath34 we are having @xmath42 . for @xmath38 and @xmath66
they are @xmath43 and @xmath44 , respectively .
let us turn to fig .
[ fig:1]b and write down the equations of contact between the cylinders .
vector @xmath67 connecting a point on one cylinder axis with a point of the other cylinder axis , can be written as : @xmath68 where again @xmath57 and @xmath69 are the affine parameters along the cylinder axes .
as far as at the touching point @xmath70 being the shortest vector should be orthogonal to both axes of the cylinders @xmath8 and @xmath14 , one can write down two equations to pin - point the place of touching which determine @xmath57 and @xmath69 : @xmath71 then for @xmath57 and @xmath69 one obtains from eq .
( [ eq:9 ] ) : @xmath72 @xmath73 on substituting eqs .
( [ eq:10 ] ) and ( [ eq:11 ] ) into eq .
( [ eq:8 ] ) one finally gets the first equation to solve : @xmath74 with all the rest equations being obtained by replacing @xmath75,@xmath76,@xmath77,@xmath78 , where @xmath79,@xmath80 and @xmath81
. the solving block that in mathcad begins with `` given '' and ends with `` find '' contains either @xmath42 , @xmath43 , or @xmath44 equations for 7- , 8- , and 9-knots , correspondingly , in accordance with eq .
( [ eq:7 ] ) .
the accuracy of the mathcad calculations was @xmath1 in order .
the equations were calculated for angles and sometimes for radii when it was appropriate . in case of @xmath34
the numbers of unknowns was @xmath82 , thus coinciding with the number of the equations , because we fixed @xmath0 ( any angle from @xmath7 to @xmath11 ) , @xmath63 , and @xmath62 . in case of @xmath38 the number of unknowns was @xmath83 because we now calculated also for @xmath0 , remaining @xmath63 , and @xmath62 fixed , and also two of the radii were taken as unknowns and calculated , for example @xmath84 and @xmath85 as in fig . [ fig:8]a below , or @xmath86 and @xmath85 in fig .
[ fig:8]b . in case of @xmath40
the number of unknowns was @xmath87 , where we added @xmath31 radii @xmath88,@xmath84,@xmath86,@xmath89,@xmath85,@xmath90 to unknowns .
we could additionally control the calculations and topology of configurations by utilizing the possibility to calculate @xmath91 also analytically as it was performed in [ 2 ] , [ 3 ] .
let us fix @xmath62 .
it is easy to make sure from eqs .
( [ eq:5 ] ) , ( [ eq:10 ] ) , ( [ eq:11 ] ) that @xmath70 of eq .
( [ eq:8 ] ) is a linear function of @xmath92 . after rewriting eq .
( [ eq:8 ] ) as @xmath93 from eq .
( [ eq:13 ] ) one obtains by inserting @xmath61 either @xmath7 or @xmath8 : @xmath94 and @xmath95 . substituting eq .
( [ eq:13 ] ) into eq .
( [ eq:12 ] ) one obtains a quadratic equation for @xmath61 @xmath96 which solution is : @xmath97 ( \vec \lambda_{12 } ) ^2 } } { ( \vec \lambda_{12 } ) ^2 } .
\label{eq:15}\ ] ] here @xmath98 defines the sign of the square root to be either @xmath99 or @xmath100 and discriminates the topology .
replacing index @xmath101 gives the solutions for @xmath102.the choice of topology for 7-knot in fig .
[ fig:2 ] corresponds to @xmath100 ; @xmath103 ; @xmath104 ; @xmath105 ; @xmath106 . for 7 * -knot of fig .
[ fig:3 ] the topology corresponds to @xmath99 ; @xmath107 ; @xmath104 ; @xmath108 ; @xmath109 , the same set as in @xcite . with the help of eq . ( [ eq:15 ] )
we can compare the precision of the calculations .
for example , for 7-knot in fig .
[ fig:2 ] we obtained ( setting the tolerance to zero at @xmath32 ) @xmath91 in two ways : once calculating with full @xmath42 equations similar to eq .
( [ eq:12 ] ) for all @xmath42 @xmath110,@xmath111,@xmath91 ( the second column in table [ tab:1 ] ) and then re - calculating @xmath91 analytically from eq .
( [ eq:15 ] ) ( the third column in table [ tab:1 ] ) .
one can see rather good accuracy for both 7-knot and 7 * -knot .
llllll 7-knot & calc .
( 12 ) & calc . with eq .
( 15 ) & 7 * -knot & calc .
( 12 ) & calc . with eq .
( 15 ) + @xmath61 & -22.893487139783 & -22.893487139783 & @xmath61 & 4.443443932143 & 4.443443932142 + @xmath112 & -16.244288714448 & -16.244288714448 & @xmath112 & -6.499245991778 & -6.499245991776 + @xmath113 & 24.121492295218 & 24.121492295218 & @xmath113 & 2.031253999672 & 2.031253999672 + @xmath114 & -35.355106987103 & -35.355106987102 & @xmath114 & -3.139447612434 & -3.139447612434 + @xmath115 & 4.208529446227 & 4.208529446227 & @xmath115 & -3.866166673959 & -3.866166673959 + the main challenge in the calculations for @xmath34 was first to locate the appropriate initial values for @xmath110,@xmath111,@xmath91,@xmath116 to find any solution possible which was realized in @xcite ( fig .
[ eq:2 ] ) .
the second challenge was to find the topology that provides the most entangled structure with only three `` unknotted '' cylinders ( the cylinders any of which can be translated to infinity without disturbing all the rest , see below ) like the three cylinders , marked @xmath31(violet ) , @xmath23(green ) , @xmath14(blue ) in 7 * -knot with all cylinders equal , in fig .
[ fig:3]b .
understanding that having the tightest configuration might be important for looking for configurations with @xmath117 stipulated our success in finding firstly 8-knot and then the ultimate 9-knot .
let us describe the results obtained .
as it is seen from eqs .
( [ eq:3 ] ) and ( [ eq:4 ] ) , @xmath37 for @xmath34 .
this degree of freedom is controlled by the angle @xmath0 which can be sweeping from @xmath7 to @xmath11 while the topology of a configuration does not change .
however , the topology is connected with the set of radii . for example , one can have the 7-knot from fig .
[ fig:2 ] with the radii @xmath118 as well as the 7 * -knot ( tilde indicates the same topology but not all equivalent radii ) from fig . [ fig:3 ] with the same set @xmath119 , so that @xmath120 for both cases .
let us vary only @xmath86 .
then , while for 7 * -knot @xmath86 can be reduced to @xmath8 , in case of 7-knot @xmath86 can be reduced only to @xmath121 , where some of the cylinders try to get parallel . on the other hand , 7
* -knot topology configuration can not exist for @xmath122 as the calculations show ( see fig .
[ fig:6]a for @xmath123 as the cylinders try to get parallel ) .
a traverse in the @xmath124 space of radii set thus switches the configuration to different topology .
topology is preserved in radii variations until the cylinders become parallel .
there should exist non - overlapping sets of radii ( `` basins '' or `` valleys '' ) that have one - to - one correspondence with the topology .
at least for the vicinity of the set of the radii @xmath125 it is true : this set selects 7 * -knot topology uniquely .
let us show quite simply that the topology of the two configurations in fig .
[ fig:2 ] and fig .
[ fig:3 ] is different with the help of the notions of `` knotted '' and `` unknotted '' cylinders , illustrated in fig .
[ fig:4 ] on example of two projections of 7 * -knot .
indeed , it is impossible to remove the 0th cylinder by translating it in any direction orthogonal to its axis without disturbing the rest in fig . [ fig:4]a . on contrary
, the 2nd cylinder can be translated to infinity in the direction of the black arrow in fig .
[ fig:4]b .
one can see by inspection for the 7-knot in fig .
[ fig:2 ] that there are four unknotted cylinders @xmath14 , @xmath23 , @xmath10 , @xmath28 and for 7 * -knot in fig .
[ fig:3 ] there are only three unknotted cylinders @xmath14 , @xmath23 , @xmath31 ; therefore the configurations are topologically different .
note that even if we `` relax '' 7-knot configurations by translating the cylinders arbitrarily in directions orthogonal to their axes ( so that they stop touching each other ) , but without changing their orientations , the configurations would retain their entanglement and topology as far as the topology can change only through the parallelism of cylinders .
it indicates that to any configuration of infinite cylinders some numerical characteristics can be ascribed which is in one to one correspondence with the topology of the configuration .
we suggest the simplest symmetric matrix that discerns the topology with a typical normalized element of chirality : @xmath126 which renders either @xmath127 or @xmath128 , and we put zero on the diagonal . for the orientation of vectors @xmath48 and
@xmath129 being defined with respect to the pivotal cylinder as shown in fig .
[ fig:1 ] , and vector @xmath67 being orthogonal to both direction while satisfying eq .
( [ eq:8 ] ) , any change from @xmath127 to @xmath128 may occur either through the parallel configuration ( which is a degeneracy ) or through the mirror reflection ( which is a discreet operation ) .
thus , chirality matrix @xmath130 of eq .
( [ eq:16 ] ) is preserved when the topology of the entanglement of the cylinders is preserved , and it makes the classification of entangled configurations of straight cylinders much simpler than classification of conventional one - component knots . for 7-knot from fig . [ fig:2 ] , the matrix @xmath130 is calculated to be @xmath131 with its determinant @xmath132 . for 7 * -knot given in fig .
[ fig:3 ] , the matrix @xmath130 is obviously different @xmath133 with its determinant @xmath134 .
we made sure that the matrix @xmath130 remains invariant while the scissor angle @xmath0 sweeps the whole range from @xmath7 to @xmath11 .
the first row and the first column always contain only @xmath127 because of the chosen right - hand orientation way for angles @xmath110 and @xmath111 ( fig .
[ fig:1]a ) at the contact with the pivotal 0th cylinder . given
this matrix one can unambiguously interweave 7 cylinders in a 7-knot .
there are also interesting properties of the matrix of eq .
( [ eq:16 ] ) with non - normalized elements @xmath135 .
7-knots may be connected with 7-tangles with some reservations . indeed , according to conway s definition ,
an n - tangle is a proper embedding of a disjoint collection of @xmath136 arcs into a 3-ball when the embedding must send the endpoints of the arcs to 2n marked points on the ball s boundary . for the case of 7-knot
the endpoints are strictly polar points of the ball ( fig .
5a ) where the position of @xmath137 points is dictated by the radii and the conditions of touching .
yet the arcs , equipped with non - zero thickness while representing the touching cylinders , should also be pairwise cross - linked at one point of crossing .
this cross - link makes it impossible for a third arc to slip `` between '' the two arcs , thus producing a `` knotting '' in the 7-tangle .
let us illustrate it by a projection diagram in fig .
[ fig:5]b .
it is known that without loss of generality , one can consider the marked points on the 3-ball boundary to lie on a great circle .
the tangle can be arranged to be in general position with respect to the projection onto the flat disc bounded by the great circle which is the tangle diagram shown in fig .
[ fig:5]b .
the inset to this figure illustrates in detail the attempt to re - deform the 4th cyan arc into the one shown with the dotted line which lies along the great circle and symbolizes the movement of the corresponding cylinder to infinity .
it is impossible because at the position of the cross - link ( indicated by the black arrow ) between the 2nd ( blue ) and the 3rd ( green ) arcs , the 4th arc should `` slip '' between the 2nd and the 3rd arcs , which is prevented by their cross - link ( see the caption to fig . [
fig:5]b ) . on contrary
, the 6th ( violet ) arc can be re - deformed in the north - west direction to lie along the great circle .
the 6th arc , while passing through the cross - link marked by the red arrow , never gets between the 0th ( red ) and the 3rd ( green ) arcs .
therefore , in consistency with above - said , the 4th arc ( as well as the 4th cylinder ) is `` knotted '' and the 6th arc ( as well as the 6th cylinder ) is `` unknotted '' . the 7-tangle diagram with cross - links indeed reflects the topology of the contacting cylinders in the 7 * -knot .
let us give some computational results for 7 * -knot ( the tolerance to zero was set @xmath138 ) . as we mentioned above
, the scissor angle @xmath0 between the 0th pivotal cylinder ( red ) and the 1st cylinder ( brown ) varies from @xmath7 to @xmath11 . in fig .
[ fig:6]c we give a side view of the configuration where @xmath0 approaches @xmath7 and the cylinders get parallel .
[ fig:6]d gives the computed dependence of all latitudinal angles @xmath110 vs @xmath0 .
one can notice that the 6th and the 5th cylinders do not upend from @xmath7 to @xmath11 or vise versa .
being projected along the pivotal cylinder the 7 * -knot demonstrates two limiting projections : one in fig .
[ fig:7]a corresponds to @xmath139 and the other in fig .
[ fig:7]b corresponds to @xmath140 .
the total computed dependencies of @xmath111 vs @xmath0 are given in fig .
[ fig:7]c . grows larger some cylinders tend to become parallel ; ( b ) when @xmath86 grows smaller some cylinders also tend to become parallel ; ( c ) for 7 * -knot as @xmath139 cylinders grow parallel , the same holds for @xmath140 ( see online resource 1 animation ) ; ( d ) the graph demonstrates the dependence of latitudinal angles @xmath110 on the scissor angle @xmath0 for 7 * -knot .
the curves are attracted either to @xmath7 or to @xmath11 as @xmath0 approaches either @xmath7 or @xmath11 ( the cylinders get parallel),scaledwidth=85.0% ] and ( b ) @xmath140 ; ( c ) the graph demonstrates the dependence of longitudinal angles @xmath36 on the scissor angle @xmath0 .
the blue arrows indicate the limiting angles @xmath111 for configurations in ( a ) and ( b ) , scaledwidth=85.0% ]
although it is possible to calculate 8-knot with changeable nonequivalent radii we present in fig .
[ fig:8 ] only two unique configurations with calculated ( @xmath141 ; @xmath142 ) and ( @xmath143 ; @xmath144 ) while all the rest @xmath31 cylinders in the 8-knot are of the unit radii .
any of such configurations with two determined radii and all the rest of the unit size is unique , being @xmath44 configurations in total if one combines all possible pairs of radii .
they can be calculated by using the mathcad program text given in online resource 4 .
the 8-tangle for this configuration can be drawn as well .
the control of the topology of the configuration of the cylinders according to their degree of knotting is the same as for 7-knot .
for example , exactly like in the 7-knot with equal cylinders only three cylinders 6th ( violet ) , 3rd ( green ) , and 2nd ( blue ) are unknotted . finally , the matrix @xmath130 for both configurations in fig . [ fig:8 ] is @xmath145 and @xmath85 are chosen ; ( b ) @xmath86 and @xmath85 are chosen , scaledwidth=85.0% ] with its determinant @xmath146 .
we introduced the upper script @xmath147 to indicate the 8-knot .
one can notice that the left upper corner @xmath148 block of the matrix exactly coincides with the matrix in eq .
[ eq:18 ] . indeed , because we constructed the 8-knot by making use of the tightest 7 * -knot configuration to insert the eighth cylinder , this topology remained `` frozen '' .
this one is more interesting because it is the ultimate configuration possible for the infinite round cylinders .
it closely resembles a complicated version of apollonius circles in the respect that three radii determine all the rest .
we present two of 9-knot configurations of the same topology in fig .
[ fig:9 ] .
the one in fig .
[ fig:9]a was the first one ever as obtained at ( @xmath3 ; @xmath149 ; @xmath150 ) .
calculated radii are ( @xmath151 ; @xmath152 ; @xmath153 ; @xmath154 ; @xmath155 ; @xmath156 ) .
the other configuration in fig .
[ fig:9]b is obtained at ( @xmath3 ; @xmath157;@xmath158 ) with calculated radii ( @xmath159 ; @xmath160 ; @xmath161 ; @xmath162 ; @xmath163 ; @xmath164 ) .
the latter configuration is unique because it has the least ratio of @xmath165 between the largest and the smallest radii in the radii set of all the @xmath166 sets calculated .
the point ( @xmath157;@xmath167 ) which corresponds to this unique configuration is marked with a star inside the domain of the allowed radii @xmath47 and @xmath168 that we explored and depicted in fig .
[ fig:9]c . in principle
, the rest @xmath31 radii can be visualized as @xmath31 surfaces roofing this domain .
the boundary points of the domain may include configurations where the radius of at least one of the cylinder either expands to infinity ( mostly the upper part of the domain as shown in fig .
[ fig:9]d ) or shrinks to zero ( mostly the lower part of the domain ) . ,
@xmath168 ) for existing configurations in double - log scale .
587 dots mark existing configurations .
the configuration ( b ) is marked with the star ; ( d ) one of the configurations which is near the border of the domain showing the tendency of one radius to infinity ( the part of the huge cylinder looking like a ring ) with some others being nearly parallel , scaledwidth=85.0% ] the topological classification of 9-knot is possible in analogy with 7-knot discussed above .
for example , exactly like in the 7-knot with equal cylinders only three cylinders : 6th ( violet ) , 3rd ( green ) , and 2nd ( blue ) are unknotted for 9-knot .
the most knotted `` inner '' cylinders can be released only after removing @xmath14 others one by one . for the matrix @xmath169 we obtained for the three configurations shown in fig .
[ fig:9 ] @xmath170 with its determinant @xmath171
. this degeneracy may point out to some additional symmetry hidden in this ultimate configuration of 9-knot .
we did not check for all @xmath166 configurations calculated for the outline of the domain in fig . 9c but where we checked the matrix was the same .
thus we conclude that the topology was likely to be preserved .
one may notice that the left upper corner @xmath172 block of the matrix exactly coincides with eq .
( [ eq:18 ] ) which was the consequence of constructing the 9-knot by introducing the ninth cylinder into the previous 8-knot configuration and then recalculating .
it is for now an open question whether other topologies exist for 9-knot .
we have given only a glimpse of a rich world of geometry of cylinder configurations that is to be revealed in detail .
numerous generalizations and questions come to mind , such as : what if the cylinders are having arbitrary cross - sections : elliptic , polygonal , or just a segment , a cylinder being degenerated into a flat band ( excluding though the degenerate solutions of more than two edges touching in just one point ) ?
what if such bands are not flat , what if the axes of the cylinders are arbitrary curves that still connect `` one infinity '' with the other one , etc . ?
with the help of our method of calculating the degrees of freedom it is possible to estimate the relevant numbers but there should be advanced analytics or computations to make sure whether an arrangement is realizable like it happened in case of 7- , 8- , and 9-knots . in our previous works it was already shown that entanglement of the cylinders brings unconventional mechanical ( auxetic ) properties to the ensembles of cylinders by using 6-knots as building blocks .
now we introduced a matrix that identifies the topology of cylinder configurations and probably may be used for a description of ensembles of entangled cylinders or rods that goes beyond the usual onsager approximation of low packing density and pairwise interaction of rigid rods . on this way one may expect more complicated entangled and thermodynamically stable structures than the well - known nematic ordering of rigid rods to exist , for example , similar to spin - glasses .
many objects in the field theory such as strings and vortexes have the geometry of cylinders .
the difference between the cylinder and the sphere may resemble the difference between particles with spin and spinless particles .
one may notice an intriguing parallel between the geometry of infinite cylinders , punching the `` celestial sphere '' , and twistor geometry of light rays @xcite .
a light ray in this picture lies along the axis which is not structureless but is equipped with the information of the helicity ( chirality ) .
contacting cylinders and other cylinder - like geometrical constructions may provide a complementary picture for twistors . note that as we showed above just a configuration of two cylinders produces chirality similar to the helicity of the twistor @xcite .
we found here that the topology of entanglement of cylinders can be described in easier ( and one to one ) manner than for conventional one - component knots .
this finding may be useful for the information theory as well .
we anticipate that some confusion with the terminology may occur in future .
there has been a term of n - knot which is a single n - dimensional sphere @xmath173 embedded in m - dimensional sphere @xmath174 .
therefore it might be needed to alter our term `` n - knot '' for a configuration of n mutually touching cylinders for something that we may suggest to be `` n - cross '' .
pikhitsa , a simple model to evaluate the percolation threshold in a continuum problem , sov .
solid state * 28 * , 1220 - 1222 ( 1986 ) .
pikhitsa , regular network of contacting cylinders with implications for materials with negative poisson ratios phys .
* 93 * , 015505(1 - 4 ) ( 2004 ) .
pikhitsa , m. choi , h .- j .
kim , and s .- h .
ahn , auxetic lattice of multipods , phys stat solidi b * 246 * , 2098 - 2101 ( 2009 ) .
s. bozoki , t .- l .
lee , and l. ronyai , arxiv:1308.5164 ( 2013 ) .
r. penrose , the road to reality , jonathan cape , london , ( 2004 ) . | it has been a challenge to make seven straight round cylinders mutually touch before our now 10-year old discovery [ phys .
rev . lett . * 93 * , 015505 ( 2004 ) ] of configurations of seven mutually touching infinitely long round cylinders ( then coined 7-knots ) .
because of the current interest in string - like objects and entanglement which occur in many fields of physics it is useful to find a simple way to treat ensembles of straight infinite cylinders . here
we propose a treatment with a chirality matrix . by comparing 7-knot with variable radii with the one where all cylinders are of equal radii ( here 7 * -knot , which for the first time appeared in [ phys .
stat .
solidi , b * 246 * , 2098 ( 2009 ) ] ) , we show that the reduction of 7-knot with a set of non - equal cylinder radii to 7 * -knot of equal radii is possible only for one topologically unique configuration , all other 7-knots being of different topology .
we found novel configurations for mutually touching infinitely long round cylinders when their numbers are eight and ultimately nine ( here coined 8-knots and 9-knots ) . unlike the case of 7-knot , where one angular parameter ( for a given set of fixed radii ) may change by sweeping a scissor angle between two chosen cylinders , in case of 8- and 9-knots their degrees of freedom are completely exhausted by mutual touching so that their configurations are `` frozen '' for
each given set of radii .
for 8-knot the radii of any six cylinders may be changeable ( for example , all taken equal ) while two remaining are uniquely determined by the others .
we show that 9-knot makes the ultimate configuration where only three cylinders can have changeable radii and the remaining six are determined by the three .
possible generalizations and connection with physics are mentioned .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
to verify the proper transmission of squeezed states we show a measurement of the squeezed observable @xmath16 over a wide transmission range in the main letter ( see fig . 2 therein ) . additionally , we measured the stokes observable @xmath28 and show the result in fig .
[ fig : abschw_antisq ] . as mentioned in the main letter , this observable exhibts antisqueezing , but is dominated by classical excess noise from photon - phonon interaction in the fiber that we use for squeezed light generation .
the variance of @xmath28 ( red crosses ) scales quadratically with the transmission value ( fitted in yellow ) , while the black line indicates the shot noise limit as given by the fitted curve of fig . 2 of the main letter .
+ in the main letter we discuss the reconstruction of the density matrix of a transmitted squeezed state and show the wigner function , which is calculated from that density matrix in fig .
we directly show the essential part of the density matrix ( see fig .
[ fig : rho ] ) .
the density matrix is dominated by the contributions on the diagonal , but an important signature of squeezing can be seen in the non - zero off - diagonal matrix elements .
if only matrix elements of the diagonal would contribute to the density matrix , the resulting wigner function would be radially symmetric . during the squeezing process
correlated pairs of photons are generated in the sidebands aside of the signal carrier .
these correlations manifest themselves in contributions at matrix elements with both even indices ( e.g. @xmath37 ) . in the reconstructed density matrix however , also matrix elements with
both odd indices ( e.g. @xmath38 ) are contributing to the state .
this is due to ( classical ) excess noise . | continuous variable quantum states of light are used in quantum information protocols and quantum metrology and known to degrade with loss and added noise . we were able to show the distribution of bright polarization squeezed quantum states of light through an urban free - space channel of 1.6 km length . to measure the squeezed states in this extreme environment ,
we utilize polarization encoding and a postselection protocol that is taking into account classical side information stemming from the distribution of transmission values .
the successful distribution of continuous variable squeezed states is accentuated by a quantum state tomography , allowing for determining the purity of the state .
the survival of quantum states in hostile environments is of crucial importance in quantum information processing and of general interest in the understanding of decoherence processes .
the transmission of quantum states of light through an atmospheric link is such a demanding task .
free - space quantum state distribution has the potential to form a key component in future quantum networks .
it offers the possibility of distributing quantum states with high flexibility in the required infrastructure .
furthermore , links to moving objects , like vehicles , planes , and satellites are conceivable .
recent experiments demonstrated free - space quantum channels in long range ground - to - ground links @xcite as well as in a link from a ground station to a plane @xcite .
other experiments study the feasibility of quantum communication between earth and space , in general @xcite , or the effects of atmospheric turbulence on quantum states , in particular , and how to exploit them @xcite . all these experiments ( for a general review on free - space quantum communication experiments see ref .
@xcite ) , are based on discrete variables , i.e. , rely on photon counting at the receiver . on the other hand ,
recent investigations have proven the feasibility of continuous variable ( cv ) free - space quantum key distribution @xcite . by using homodyne receivers ,
this approach can offer high speed detection and immunity to stray light .
a comprehensive review on cv gaussian quantum information can be found in ref .
@xcite .
squeezed states can be used in cv quantum key distribution protocols @xcite and it has been proven @xcite that they can provide an enhancement compared to coherent states . when transmitting cv quantum states much care has to be taken to avoid their decoherence .
losses will result in a degradation of the quantum states , whereas in discrete variable systems , these will merely lower the detection rate . also , phase relations and wave front distortions play an important role , as the applied homodyne detection relies on optical interference and a high homodyne efficiency is crucial .
thus , the safe transmission of cv quantum states in harsh environments poses a major challenge in different physical systems , ranging from optical free - space links to superconducting cavity quantum electrodynamics @xcite . in this letter
, we demonstrate the distribution of bright cv squeezed states of light through a turbulent atmospheric channel .
the atmosphere introduces fluctuations in amplitude and phase . because of the fluctuating phase and the resulting wave front distortions , standard homodyne measurements at the receiver are challenging .
we circumvent this problem by using polarization encoding .
polarization offers a distinct advantage in this case .
the different polarization components can be measured separately , and yet they occupy the same spatial mode and experience identical distortions . measuring the polarization components averaged over the beam cross section is like a differential measurement , where the distortions cancel out . in that sense ,
the polarization measurement is analogous to measuring in a decoherence free subspace @xcite and is immune to phase fluctuations .
bright polarization squeezed states are equivalent to combining a bright coherent polarization mode and an orthogonally polarized vacuum squeezed mode @xcite .
thus , we are able to perform polarization measurements , which can be approximated as homodyne measurements with a bright local oscillator and a polarization multiplexed squeezed vacuum signal . in principle , it would be possible to use a different degree of freedom for the multiplexing of the squeezed signal with the local oscillator or even to use an independent laser at the receiver as a local oscillator .
however , it would be technically challenging to achieve a high interferometric visibility .
cv polarization encoding also perfectly fits atmospheric transmission systems @xcite , and the negligible impact of the atmosphere on the state of polarization was even investigated in a space to ground scenario @xcite .
minor effects occur for the case of ice crystals in the atmosphere , which was not relevant for the urban link measurements reported in this letter . additionally , the polarization of the light field is easy to manipulate without introducing considerable loss .
homodyne receivers are , in general , immune to stray light , allowing for daylight operation without any further spectral or spatial filtering
. the atmospheric channel introduces fluctuations of the intensity through beam wandering , distortion , and spreading .
special care is necessary to take into account the resulting variation of the shot noise level during our measurements .
we developed a postselection protocol that is also able to enhance the degree of squeezing compared to that of the raw channel data .
these results are substantiated by a quantum state reconstruction via a maximum likelihood algorithm @xcite .
our experiment is based on a realistic free - space channel of 1.6 km length in the city of erlangen [ see fig . [
fig : setup_channel](a ) ] .
the channel is located in a truly urban environment .
the optical beam path starting from the sender _ ( alice ) _ runs past buildings , heated rooftops , heavy traffic , and forests .
this inhomogeneous environment leads to significant turbulence , affecting phase and intensity .
the beam is received by bob , located in the 12th floor of a university building . in the following paragraph
, we explain the details of our setup consisting of a robust source of polarization squeezed states of light at alice and a detection stage at bob . at alice , we generate cv polarization squeezed states in a compact and stable setup by exploiting the kerr nonlinearity of a single mode fiber @xcite [ see fig . [
fig : setup_channel](b ) ] .
the source is able to operate without any need of controlled laboratory conditions .
all components are built onto a breadboard with an area less than 0.3m@xmath0 .
a commercial soliton laser ( origami , onefive gmbh ) emits femtosecond pulses ( @xmath1fs ) at a center wavelength of @xmath2 nm and a repetition rate of @xmath3mhz .
shorter wavelengths offer an enhanced collection efficiency after channel transmission at a given aperture size .
therefore , the sending and receiving optics of our free - space channel are optimized for wavelengths in the 800 nm range , which also coincides with low atmospheric losses
. consequently , we use a periodically poled lithium - niobate crystal ( mshg1550 - 0.5 - 0.3 , _ covesion ltd . _ ) to convert the output of the laser to its second harmonic at @xmath4 nm and dichroic mirrors for subsequent spectral filtering . for the preparation of the cv polarization squeezed states we use both axes of a birefringent polarization maintaining photonic crystal fiber ( nl - pm-700 , zero dispersion wavelength @xmath5 nm , crystal fibre a / s ) . because of the nonlinear kerr effect in this fiber , we individually quadrature squeeze two orthogonally polarized pulses .
these two pulses are overlapped to produce a circularly polarized state , showing polarization squeezing . to achieve this
, we have to precompensate the birefringence of the fiber with different optical path lengths in front of the fiber .
we actively control the overlap and the phase between the two pulses by tapping 0.1% of the signal after the fiber and controlling a piezoelectric actuator used in an interferometerlike configuration before the fiber . for the description of the state of polarization the stokes operators @xcite
are used : @xmath6 with the field operators for the horizontally and vertically polarized modes @xmath7 and @xmath8 . in terms of these stokes operators , our state
has a mean value of 0 for @xmath9 and @xmath10 , while @xmath11 . in poincar
space , the plane perpendicular to @xmath12 is called a dark plane in this case .
the combined circularly polarized light pulses have a continuous wave equivalent optical power of @xmath13mw and show squeezing along one particular measurement direction within the @xmath14-@xmath15 dark plane @xmath16 and antisqueezing in the orthogonal direction @xmath17 @xcite .
we measured the squeezing value at alice with an electrical spectrum analyzer at a sideband frequency of 12mhz , a resolution bandwidth of 300khz , and a video bandwidth of 30hz .
for the measurements at bob reported in this letter , we measured a value of squeezing of @xmath18db at alice . the high antisqueezing value of @xmath19db is a result of quantum noise and excess noise from photon - phonon interactions in the fiber .
finally , the optical beam is expanded by a telescope decreasing the divergence of the beam .
the prepared states are distributed through the free - space channel . at bob , it is important to collect as much as possible of the incoming light field to minimize degrading loss .
simultaneously , it is crucial to avoid any disturbance of the state of polarization .
therefore , we do not use a mirror but an achromatic lens with a diameter of 150 mm and a focal length of @xmath20 mm to collect the light .
a half - wave plate ( hwp ) and a polarizing beam splitter ( pbs ) are placed slightly before the focal point , such that the beam can be directly focused onto two pin - diode ( s3883 , quantum efficiency @xmath21 , _ hamamatsu photonics k.k .
_ ) detectors behind the pbs [ see fig .
[ fig : setup_channel](c ) ]
. the ac outputs of these custom made detectors are mixed with a 12mhz radio frequency local oscillator ( arbitrary waveform generator 33250a , _ agilent technologies inc .
_ ) , amplified ( wideband voltage amplifier dhpva-100 , _ femto messtechnik gmbh _ ) , low pass filtered ( blp-1.9 , _ mini - circuits laboratory _ ,
3db cutoff frequency 2.5mhz ) , and digitized by an analogue - to - digital converter ( ad card , compuscope cs1610 , _ gage a product brand of dynamicsignals llc _ ) with a sampling rate of 10msamples / s .
the bandwidth of the dc outputs , that are sampled simultaneously with identical sampling rates is @xmath22khz and thus sufficient for the atmospheric fluctuations .
this setup allows us to measure the stokes observables in the @xmath14-@xmath15 dark plane at a sideband frequency of 12mhz with a bandwidth of 2.5mhz ( given by the used low pass filters ) by calculating the difference of the two , still oversampled , ac signals .
the measured stokes observable within the @xmath14-@xmath15 dark plane can be chosen by rotating the hwp .
the sum of the ac signals gives rise to the @xmath23 signal .
the sum of the dc signals , from which we can calculate the respective transmission values , can be used as classical side information .
-@xmath15 dark plane , dependent on the orientation of the hwp . ]
a turbulent atmosphere leads to beam wandering , distortion , and spreading and thus to continuous fluctuations of the intensity at the receiver .
this directly results in a varying shot noise level during our measurements .
it is important to account for this with a proper shot noise calibration and monitoring .
we use the transmission value obtained as classical side information and sort all collected data according to the prevailing transmission value into bins with a width of @xmath24% of the total optical power sent from alice ( corresponding to @xmath25@xmath26w ) . with this technique , we are able to compare the measured stokes parameter variance to the corresponding shot noise variance . a reliable way to prove genuine detection of squeezed states is to compare the attenuation of the quantum states with its theoretical prediction . squeezed states show a distinct behavior that can be discriminated from classical noise and rule out technical effects such as saturation of the detectors .
we use both the natural fluctuation in transmission and additional attenuation of the beam imposed at the receiver to be able to measure the squeezed states at bob for all transmission values down to 0 ( see fig . [
fig : abschw_sq ] ) .
the additional attenuation was achieved by moving an optically dense material into the beam at the receiver stage .
we carefully adjust the hwp to measure @xmath16 ( @xmath17 ) by minimizing ( maximizing ) the measured variance .
after the measurement we average the variances of the total intensity and the stokes parameters over 10000 samples for each individual transmission bin and discard all bins containing fewer than 50000 samples .
thus , at least five variances and their mean value were obtained for each bin . to give a conservative estimate of the error of these measurements
the error bar indicates 1 standard deviation of the deduced variances .
the @xmath23 signal ( blue circles ) scales linearly with the transmission ( black fit ) , which is in perfect agreement to a shot noise limited signal .
the polarization squeezed signal @xmath27 ( red crosses ) scales quadratically ( yellow fit ) , as expected from theory @xcite , and is below the shot noise level .
the normalized green histogram visualizes the transmission statistics during this measurement .
signal , which determines the shot noise limit , scales linearly with the transmission value ( fitted in black ) .
in good agreement to theory the polarization squeezed signal @xmath16 scales quadratically with the transmission value ( fitted in yellow ) . we show only every third data point for clarity .
the green histogram shows the transmission statistics , which is a result of the natural fading channel and artificial attenuation . ]
the measurement of the stokes observable @xmath28 ( see the supplemental material @xcite ) shows antisqueezing as well as classical excess noise stemming mostly from photon - phonon interaction in the fiber used for squeezed light generation in the sender setup . after verifying the genuine transmission of squeezed states , we investigated the natural transmission statistics of our free - space fading channel , which is mainly influenced by the clipping of the beam at the receiving aperture ( the absorption in the atmosphere in clear weather conditions typically amounts to -0.1db / km @xcite , which contributes less than 4% for a link of 1.6 km length ) .
the beam spot size of the received beam is broadened due to the wave front distortions originating from the atmospheric influence .
additionally , the atmospheric turbulence leads to beam wandering .
the transmission statistics is shown in fig .
[ fig : sq ] as a normalized green histogram .
the shot noise variance ( black line ) is determined by the corresponding measurement shown in fig .
[ fig : abschw_sq ] .
the squeezing value for the squeezed state with the highest transmission value is @xmath29db below the shot noise variance , slightly increased compared to the average amount of squeezing of @xmath30db .
both values are obtained without any correction for losses .
the enhancement by the postselection protocol which selects the transmission bin with the highest value depends on several parameters .
this will be more pronounced if the investigated channel transmission covers a broader range or if the highest possible channel transmission would be closer to the physical limit of 1 .
finally , this postselection technique becomes most important if the initial squeezing is very high , as squeezing degrades in a highly nonlinear fashion with respect to losses .
db to @xmath31db relative to shot noise variance . ]
the state selection protocol enables us to perform a quantum state reconstruction of a selected squeezed state after the noisy channel . here
, we can interpret the @xmath14-@xmath15 dark plane as the quadrature phase space as @xmath32 .
we select the state with 55.2% transmission .
we chose this state because it offers a sufficiently good statistic for all measured angles .
we measure the stokes observables at 29 different angles between the squeezing direction @xmath16 and the antisqueezing direction @xmath17 . at each angle
, we use a total of @xmath33 samples which are subsequently sorted into 251 bins to derive the corresponding tomograms .
accounting for the anticipated symmetry of the squeezed state , we mirror the acquired tomograms to attain a full tomographic dataset comprising 58 different measurement angles in the dark plane @xcite . based on these tomograms
, we finally reconstruct the density matrix @xmath34 in the fock state basis via a maximum likelihood algorithm @xcite .
again , we do not correct the reconstructed state for losses .
the reconstruction requires truncating the density matrix to an adequate maximum photon number , where in our case a matrix with dimension 54 x 54 , i.e. with a maximum photon number of 53 , proved to be fully reconstructible from 58 tomograms @xcite .
any matrix element of higher order only contributes to the state with a relative weight of less than @xmath35 with respect to the largest matrix element , i.e. the vacuum state . from the reconstructed state
we determine a purity @xmath36 .
the essential sector of the reconstructed density matrix is shown in the supplemental material @xcite .
a more intuitive grasp of the state s properties is provided by means of the wigner function which we calculate @xcite from the reconstructed density matrix .
the result is plotted in fig .
[ fig : wigner ] . in this representation ,
the comparison between the contour lines of the coherent vacuum state and the reconstructed state directly reveals the preservation of the squeezing .
contour for the squeezed state and the vacuum state , respectively . ]
we successfully distributed polarization squeezed states through a free - space channel . to the best of our knowledge we show the first distribution of continuous variable nonclassical states through the turbulent atmosphere .
we presented a technique for taking into account a natural drifting shot noise variance .
furthermore , we enhanced the amount of squeezing of the state via a state selective protocol and reconstructed a selected state .
this substantiates the feasibility of cv free - space quantum communication and paves the way for further free - space quantum communication experiments such as squeezing distillation @xcite , cv entanglement distribution and distillation @xcite , or exploiting cv quantum states for atmospheric sensing .
the project was supported under fp7 fet proactive by the integrated project q - essence and chist - era ( hipercom ) .
the authors acknowledge support of the erlangen graduate school in advanced optical technologies ( saot ) by the german research foundation ( dfg ) in the framework of the german excellence initiative .
+ the authors thank ch .
wittmann , l. l. snchez - soto and d. sych for fruitful discussions and our colleagues at the fau computer science building for their kind support and for hosting the receiver station .
+ c. p. and b. h. contributed equally to this work .
99 r. ursin , f. tiefenbacher , t. schmitt - manderbach , h. weier , t. scheidl , m. lindenthal , b. blauensteiner , t. jennewein , j. perdigues , p. trojek , b. mer _
et al . _ , nat
. phys . * 3 * , 481 ( 2007 ) w. t. buttler , r. j. hughes , s. k. lamoreaux , g. l. morgan , j. e. nordholt , and c. g. peterson , phys .
rev .
lett . * 84 * , 5652 ( 2000 ) .
r. j. hughes , j. e. nordholt , d. derkacs , and c. g. peterson , new j. phys .
* 4 * , 43 ( 2002 ) .
t. schmitt - manderbach , h. weier , m. frst , r. ursin , f. tiefenbacher , t. scheidl , j. perdigues , z. sodnik , c. kurtsiefer , j. g. rarity , a. zeilinger , and h. weinfurter , phys .
rev .
lett . * 98 * , 010504 ( 2007 ) .
a. fedrizzi , r. ursin , t. herbst , m. nespoli , r. prevedel , t. scheidl , f. tiefenbacher , t. jennewein , and a. zeilinger , nat . phys . * 5 * , 389 ( 2009 ) .
j. yin , j .-
g .
ren , h. lu , y. cao , h .-
l .
yong , y .- p .
wu , c. liu , s .- k .
liao , f. zhou , y. jiang , x .- d .
cai _
et al . _ , nature ( london ) * 488 * , 185 ( 2012 ) .
s. nauerth , f. moll , m. rau , c. fuchs , j. horwath , s. frick , and h. weinfurter , nat . photon .
* 7 * , 382 ( 2013 ) .
j. g. rarity , p. r. tapster , p. m. gorman , and p. knight , new j. phys .
* 4 * , 82 ( 2002 ) . m. aspelmeyer , t. jennewein , m. pfennigbauer , w. leeb , and a. zeilinger , ieee j. sel . top .
quantum electron .
* 9 * , 1541 ( 2003 ) .
p. villoresi , t. jennewein , f. tamburini , m. aspelmeyer , c. bonato , r. ursin , c. pernechele , v. luceri , g. bianco , a. zeilinger , and c. barbieri , new j. phys . * 10 * , 033038 ( 2008 ) . j. m. p. armengol , b. furch , c. jacinto de matos , o. minster , l. cacciapuoti , m. pfennigbauer , m. aspelmeyer , t. jennewein , r. ursin , t. schmitt - manderbach , g. baister _
et al .
_ acta astronaut . *
63 * , 165 ( 2008 ) . c. bonato
, a. tomaello , v. da deppo , g. naletto , and p. villoresi , new j. phys . *
11 * , 045017 ( 2009 ) .
e. meyer - scott , z. yan , a. macdonald , j .-
p .
bourgoin , h. hbel , and t. jennewein , phys .
rev .
a * 84 * , 062326 ( 2011 ) .
j .-
p .
bourgoin , e. meyer - scott , b. l. higgins , b. helou , c. erven , h. hbel , b. kumar , d. hudson , i. dsouza , r. girard , r. laflamme , and t. jennewein , new j. phys . * 15 * , 023006 ( 2013 ) .
a. a. semenov , and w. vogel , phys .
rev .
a * 80 * , 021802 ( 2009 ) . c. erven , b. heim , e. meyer - scott , j. p. bourgoin , r. laflamme , g. weihs , and t. jennewein , new j. phys . *
14 * , 123018 ( 2012 ) .
i. capraro , a. tomaello , a. dallarche , f. gerlin , r. ursin , g. vallone , and p. villoresi , phys .
rev .
lett . * 109 * , 200502 ( 2012 ) .
a. tunick , t. moore , k. deacon , and r. meyers , proc .
spie * 7815 * , 781512 ( 2010 ) .
d. elser , t. bartley , b. heim , c. wittmann , d. sych , and g. leuchs , new j. phys . * 11 * , 045014 ( 2009 ) .
b. heim , d. elser , t. bartley , m. sabuncu , c. wittmann , d. sych , ch .
marquardt , and g. leuchs , appl .
phys .
b * 98 * 635 ( 2010 ) . v. c. usenko , b. heim , c. peuntinger , c. wittmann , c. marquardt , g. leuchs , and r. filip , new j. phys .
* 14 * , 093048 ( 2012 ) .
c. weedbrook , s. pirandola , r. garca - patrn , n. j. cerf , t. c. ralph , j. h. shapiro , and s. lloyd , rev .
mod .
phys . * 84 * , 621 ( 2012 ) .
m. hillery , phys .
rev . a * 61 * , 022309 ( 2000 ) .
s. lorenz , c. silberhorn , n. korolkova , r. s. windeler , and g. leuchs , appl .
phys .
b * 73 * , 855 ( 2001 ) .
v. c. usenko and r. filip , new j. phys . * 13 * , 113007 ( 2011 ) .
c eichler , d. bozyigit , c. lang , m. baur , l. steffen , j. m. fink , s. filipp , and a. wallraff , phys .
rev . lett . * 107 * , 113601 ( 2011 ) .
d. a. lidar , and k. b. whaley , _
irreversible quantum dynamics _ 83 ( springer lecture notes in physics , berlin , arxiv : quant - ph/0301032 , 2003 ) .
g. leuchs , r. dong , and d. sych , new j. phys .
* 11 * , 113040 ( 2009 ) .
j. heersink , dr . rer .
nat .
thesis , friedrich - alexander - universitt erlangen - nrnberg , 2006 .
n. korolkova , g. leuchs , r. loudon , t. c. ralph , and c. silberhorn , phys .
rev .
a * 65 * , 052306 ( 2002 ) .
m. toyoshima , h. takenaka , y. shoji , y. takayama , y. koyama , and h. kunimori , opt .
express * 17 * 22333 ( 2009 ) .
y. vardi , and d. lee , j. r. stat .
soc .
b * 55 * , 569 ( 1993 ) .
z. hradil , phys .
rev .
a * 55 * , r1561 ( 1997 ) .
a. i. lvovsky , j. opt .
b * 6 * , s556 ( 2004 ) . c. r. mller , b. stoklasa , c. peuntinger , c. gabriel , j. eh ' aek , z. hradil , a. b. klimov , g. leuchs , ch . marquardt , and l. l. s ' anchez - soto , new j. phys . * 14 * 085002 ( 2012 ) . j. heersink , v. josse , g. leuchs , and u. l. andersen , opt
. lett . * 30 * , 1192 ( 2005 ) .
j. milanovic , j. heersink , ch .
marquardt , a. huck , u. l. andersen , and g. leuchs , laser phys . *
17 * , 559 ( 2007 ) .
h. a. bachor , and t. c. ralph , _ a guide to experiments in quantum optics _ , 258 ( wiley , weinheim , 2004 ) .
see supplemental material following this letter for the figure showing the anti - squeezed stokes observable and the visualization of the reconstructed density matrix of a selected state .
v. scarani , h. bechmann - pasquinucci , n. j. cerf , m. duek , n. ltkenhaus , and m. peev , rev .
mod .
phys . * 81 * , 1301 ( 2009 ) .
d. sych , j. eh ' aek , z. hradil , g. leuchs , and l. l. s ' anchez - soto , phys .
rev .
a * 86 * , 052123 ( 2012 ) .
j. r. johansson , p. d. nation , and f. nori , comput .
. commun . * 184 * , 1234 ( 2013 ) .
j. heersink , ch .
marquardt , r. dong , r. filip , s. lorenz , g. leuchs , and u. l. andersen , phys .
rev .
lett . * 96 * , 253601 ( 2006 ) .
r. dong , m. lassen , j. heersink , ch .
marquardt , r. filip , g. leuchs , and u. l. andersen , phys .
rev .
a * 82 * , 012312 ( 2010 ) . |
accretion onto a supermassive black hole ( smbh ) is the main mechanism that sustains the powerful activity of active galactic nuclei but may also represent a common phase in the evolution of normal galaxies . a number of fundamental question about the formation of the qso phenomenon like the fuelling and triggering mechanisms are strictly related to the immediate environments of the active nucleus and in particular to its host galaxy @xcite .
smbhs may well have a period of maximum growth ( maximum nuclear luminosity ) contemporaneous with the bulk of the initial star formation in the bulge of galaxies .
studies of the co - evolution of smbh and their host spheroids are therefore obviously critical to understanding how and when galaxies in the local universe formed and evolved .
the last ten years have yielded considerable progress in characterising agn host galaxies . at variance with inactive galaxies their study
is often hampered by the presence of the luminous central source that outshines the light of the host galaxy . a problem that becomes more serious for high luminosity agns and for sources at high redshift . in spite of these limitations
the characterization of the properties of the host galaxies offers the unique opportunity to investigate the link between the central black hole mass and its host galaxy at moderate to high redshift and to trace the possible co - evolution at different cosmic epochs .
this is because for broad line agn like quasars it is possible to estimate the mass of the central bh using kinematic arguments that are not directly dependent on the host galaxies properties .
both ground - based and hst studies have shown that virtually all luminous low redshift ( [email protected] ) quasars reside in massive , spheroid - dominated host galaxies , whereas at lower luminosities quasars can also be found in early- type spiral hosts ( e.g. bahcall et al . 1997 ; dunlop et al .
2003 ; pagani et al . 2003 ; floyd et al . 2004
; jahnke et al .
this is in good agreement with the bh
bulge relationship in inactive galaxies ( e.g. gultekin et al .
2009 ) , since very massive bhs power luminous quasars .
only a small fraction of the host galaxies ( @xmath115% ) are found in merger systems but it is difficult to determine clear merger signatures from morphology alone . at low redshifts
a major contribution to the properties of quasar host galaxies has been provided by images from the hubble space telescope ( hst ) .
the improved spatial resolution has allowed the characterization of the structure and the detailed morphology of the host galaxies @xcite .
it turned out that qso are hosted in luminous galaxies that are often dominated by the spheroidal component . at high
redshift ( z @xmath21 ) hst observations of quasar host galaxies ( e.g. @xcite and references therein ) have been complemented by significant contributions from 8-m class ground - based telescopes under superb seeing conditions @xcite and/or with adaptive optics @xcite .
comparison of host galaxies of agn at high and low redshift constrain host galaxy evolution , as compared with the evolution of normal ( inactive ) galaxies .
most of the old studies of quasar host considered few tens of objects therefore in order to derive a picture of the host properties at various redshift one should combine many different samples often obtained with different telescopes and filters .
observations carried out by hst are certainly more homogeneous ( although different filters were used ) and allow to investigate a somewhat large sample based on high quality data .
nevertheless the size of these samples remain relatively small for instance in the range 0.25 @xmath3 0.5 about 50 qso were imaged by hst ( see references above ) . in order to explore a significantly larger dataset of qso
one should refer to large surveys that include both imaging and spectroscopic data . in this respect one of the most productive recent surveys
is the sloan digital sky survey that allowed to find 105783 quasars @xcite from ( dr-7 ) .
standard sdss images are , however , too shallow and the faint nebulosity around the nucleus of quasars is not detected .
this problem has been overcome in the case of the special sky region mapped by sdss for the sdss legacy survey . the central stripe in the south galactic gap , namely the stripe82 @xcite is a stripe along the celestial equator in the southern galactic cap .
it is 2.5@xmath4 wide and covers -50@xmath5ra@xmath6 + 60@xmath4 , so its total area is 275deg@xmath7 .
stripe 82 was imaged by the sdss multiple times , these data were taken in 2004 only under optimal seeing , sky brightness , and photometric conditions ( i.e. , the conditions required for imaging in the main legacy survey ; york et al .
( 2000 ) ) . in 2005 - 2007 , 219 additional imaging runs were taken on stripe 82 as part of the sdss supernova survey @xcite , designed to discover type ia supernovae at [email protected] .
the total number of images reaches @xmath1100 for the s strip and @xmath1 80 for the n strip .
the final frames were obtained by co - adding selected fields in r - band , with seeing ( as derived from 2d gaussian fit of stars and provided by sdss pipeline ) better than 2 " , sky brightness @xmath619.5 @xmath9 and less than 0.2 mag of extinction . in this area
there are 12434 quasars .
recently @xcite analyzed the stellar properties of about 800 galaxies hosting optically luminous , unobscured quasars at z@xmath0 0.6 using stripe82 images .
they focused on the color of the host galaxies and found that the quasar hosts are very blue and almost absent on the red sequence with a marked different distribution from that of normal ( inactive ) galaxies . for our study we selected qso with redshift less than 0.5 for which the stripe 82 images
allow us also to study the qso galaxy environments .
we adopt the concordance cosmology with h@xmath10 = 70 km s@xmath11 mpc@xmath11 , @xmath12 = 0.3 and @xmath13 = 0.7 . in this first paper of a series we focus on the properties of quasar hosts and their relationship with the central bh mass in the explored redshift range .
in forthcoming papers we investigate the galaxy environments @xcite and galaxy peculiarities @xcite .
a preliminary account of these results was presented in @xcite .
to derive the sample of low redshift quasars we used the fifth release of the sdss quasar catalog @xcite that is based on the sdss - dr7 data release @xcite .
it consists of qso that have a highly reliable redshift measurement and are fainter than @xmath14 , that have an absolute magnitude @xmath15 - 22 , at least one emission line with fwhm@xmath21000 km/ sec , or have complex/ interesting absorption lines .
this catalog contain @xmath1106.000 spectroscopically confirmed quasars .
our analysis is done only in the region of sky covered by the stripe82 data , these images go deeper of about @xmath12 magnitudes with respect to the usual sloan data and make possible the study of the qso hosts ( see example in fig .
[ fig : dr7_sdss82 ] ) .
we apply two main constraints on the sample .
first , we avoid objects that are closer than 0.2 deg to the edges of the stripe82 @xcite .
second , we choose an upper redshift limit z=0.5 in order to be able to resolve the quasar host for the large majority of the sample . to satisfy all these reasons we therefore select all the qsos in the range of redshift @xmath16 and in the stripe82 region i.e. @xmath17 , @xmath18 and @xmath19 .
this gives a total of 416 qsos . in this sample
we are dominated by radio quiet quasars only 24 are radio loud ( about 5% ) . in fig .
[ fig : qsosample ] we report the distribution of qso in the plane z-@xmath20 .
the mean redshift of the sample is @xmath21 = @xmath22 ( median @xmath23 ) and the average absolute magnitude is : @xmath24 = @xmath25 ( median @xmath26 ) . in table [ tab1 ] we report the main data for the qso in the sample . in column
( 1 ) i d number , in column ( 2 ) the sdss identification in columns ( 3 ) and ( 4 ) the coordinates , in column ( 5 ) the redshift , in column ( 6 ) the i band psf magnitude from sdss - dr7 , in column ( 7 ) the absolute i band magnitude , in column ( 8) the number of exposures for each co added frame , in column ( 9 ) the i band extinction and finally in column ( 10 ) the measured seeing on the co - added images .
the images used have an average seeing , as given by 2d gaussian fit of stars in the frame from sdss , of [email protected] arcsec , with a minimum of 1.01 and a maximum of 1.47 arcsec .
we have retrieved all images of the selected qso from sdss stripe 82 dataset @xcite in the i band .
this corresponds to observe in the r filter at rest frame at the average redshift of the dataset . in order to derive the properties of the galaxies hosting the qso we performed a 2d fit of the image of the qso assuming it is the superposition of two components .
the nucleus in the center and the surrounding nebulosity .
the first is described by the local point spread function of the image while for the second component we assumed a galaxy model described by a sersic law convolved with the proper psf .
the analysis of these images was performed using the astronomical image decomposition analysis ( aida , @xcite ) that was used in our previous studies of qso host galaxies @xcite .
the most critical aspect of the image decomposition is the determination of a suitable psf . in the case of sdss
images the field of view is large enough that there are always many stars in the co - added sdss image containing the target to properly derive the psf . to derive the most suitable psf of each field we have selected a number of stars ( between 5 and 15 ) in the field that are distributed around the target .
selection of these psf stars was based on various parameters as their magnitude , fwhm , ellipticity and presence of close companions . in particular ,
the selection of the stars was done according to the following criteria : the stars are not saturated ; the stars are as close to the target as possible ( while avoiding the fit region of the target ) ; the stars are as uniformly distributed around the target as possible ; the stars are sufficiently isolated ( i.e. , they have no close companions ) ; the stars cover a suitably wide range of magnitude in order to assure that the extended halo of the psf is well characterised .
we then define a radius to compute the psf model and a ring around each star where to compute the sky background .
all extra sources that were found inside these areas were masked out with an automated procedure .
the psf model was then obtained from the simultaneous fit of all selected stars using a multi function 2d model composed by 3 gaussians and one exponential function . in figure
[ fig : psf ] we show an example of the procedure adopted to derive the psf model .
it is worth to note that the psf provided by the sdss pipeline ( using source psfield ) are not adequate for this study .
this is due to a systematic underestimate of the wings of the sdss psf .
the agreement with our psf model is excellent up to @xmath1 3 arcsec from the center of the star but then a significant deviation is present ( see figure [ fig : psfconf ] ) . using the sdss psf for the qso decomposition will result in a systematic overestimate of the host galaxy luminosity and in some cases to false detection of the host galaxy signal ( see example in figure [ fig : psfconf ] panel c ) ) the second step of the analysis is to fit each quasar both with a scaled psf and with a 2 components model ( point source plus a galaxy ) .
the best fit was obtained adopting a model for the errors that include a constant term to represent the read out noise of the detector , a term representing the statistical noise due to the effective counts and an additional term that take into account the possible residual noise due to fixed pattern noise . for the coadded images we assume a readout noise of 9.5 e@xmath28 and an average gain of 3.8 e@xmath28/adu .
the term for the statistical noise is given by the coefficient 1/@xmath29 that multiply the root square of the counts .
for the residual pattern noise we assumed 2% value . .
[ fig : aida ] in figure [ fig : aida ] we show an example of the adopted procedure . in order to distinguish between resolved and unresolved objects we compared the @xmath30 of the two fits and in addition
we then inspect all the fits to further check the results .
this allows us to produce a clean list of @xmath1 350 resolved quasars by removing 7 objects ( @xmath11% of the entire sample ) that were contaminated by very bright sources in the field or defects in the image close to the targets . for 60 objects ( @xmath113% of the entire sample ) the qso were unresolved and the fit with only the psf was indistinguishable from the fit with the psf + sersic function ( see figure [ fig : chi2ratio ] ) .
the unresolved objects are mainly objects at relatively high redshift ( 42 out of 60 unresolved quasars are at z @xmath2 0.4 ) and bright nuclei . ratio of the fit ( psf+galaxy ) with respect to the one with only the psf ) .
the two vertical lines define the regions of our classification for objects resolved ( @xmath31/@xmath32 @xmath0 0.6 ) , marginally resolved and unresolved ( @xmath31/@xmath32 @xmath2 0.8 ) . ] for the unresolved objects we evaluated the bright limit of the host galaxy by adding the flux of a galaxy to the observed object until the @xmath30 of fit to these data become 20% worst of that obtained from the fit with the scaled psf .
since we have no knowledge of the underlying host galaxy we performed this evaluation of the brightness limit of the host galaxy using different models .
we assumed two type of morphology : exponential disk and de vaucoulers laws .
then we assume two values for the half light radius : 5 and 10 kpc that are representative values of the resolved objects in our sample the input half light radius in arcsec was derived assuming the redshift of the qso .
we took as upper limit to the luminosity of the host galaxy the maximum value derived from the various fit using different models and half light radius .
the final classification of the targets was based on the comparison of @xmath30 for the two fit ( only psf and psf + galaxy ) and further visual inspection of the fit . from this procedure
we classified all objects as resolved , unresolved and marginally resolved for intermediate fit ( see figure [ fig : chi2ratio ] and table [ tab : results ] ) .
in figure [ fig : mhabsr ] we show the distribution of the absolute magnitude of host galaxies for all ( 309 ) resolved quasars .
these values were corrected for the galactic extinction ( based on the values of sdss ) and k - corrected to the r band rest frame . to perform the color and @xmath33-correction transformations , we assumed an elliptical and early type spiral galaxy spectral energy distribution ( sed ) template ( @xcite ) for the host galaxy . because of the small difference of the templates in the observed spectral region , the k - correction adopting the two seds differ by few percent , thus we used for all objects the one for elliptical sed .
for the nucleus we used a composite quasar spectrum ( francis et al . 1991 ) .
all k - corrections were performed adopting these templates for the sed and convolving them with the @xmath34 and r filter responses .
an example of the procedure is illustrated in figure [ fig : kcorr ] the average absolute magnitude is @xmath35 = -22.83 @xmath27 0.6 ( median -22.86 @xmath27 0.36 ) . for comparison the distribution of absolute magnitude for a smaller ( @xmath1 100 ) sample of qso observed by hst ( see compilation by @xcite ) in the similar redshift range is @xmath35 = -23.00 @xmath27 1.05 .
the two data sets are in excellent agreement in spite of the differences of observation technique .
five quasars in our sample were observed with hubble space telescope and wfpc2 in filter f606w @xcite and it is possible to compare our analysis with the results from hst images . the comparison of the magnitudes of these host galaxies ( assuming a color correction v - r = 0.8 ) is very good ( @xmath36 = 0.1 @xmath27 0.24 ) the distribution of host galaxies in the redshift - luminosity plane ( see fig [ fig : mhz ] ) confirms previous claims that they are encompassed between m * ( m*(r ) = 21.2 ; @xcite ) and m * - 3 with more frequent distribution in the range m*-1 and m*-2 @xcite .
there is a small , but significant , increasing of the host luminosity with the redshift ( from m(r ) @xmath1 -22.5 at z @xmath10.2 to m(r ) @xmath1 -23.1 at z @xmath1 0.5 ) that is consistent with passive evolution of the underlying stellar population .
a similar trend was also reported over a wider redshift range by @xcite . while the total flux from the host galaxy is relatively well determined from the fit of resolved objects the evaluation of the half light radius is not well constrained .
this is due to some degeneracy between the effective surface brightness and effective radius that can be combined to produce the same total flux . in figure
[ fig : hlr ] we show the distribution of the effective radius ( re ) as derived from the fit to the objects with a galaxy ( modelled by a sersic law ) plus the nuclear component ( modelled by psf ) .
the average effective radius is @xmath37 = 7.7 @xmath27 3.6 kpc .
if we include only the targets with @xmath38 fit ( @xmath30 / @xmath30(psf ) @xmath0 0.5 ) and uncertainty of re @xmath0 30% the average effective radius is slightly larger : @xmath37 = 8.2 @xmath27 3.7 kpc . for the small sample of radio loud quasars ( 17 objects ) the average effective radius ( @xmath37 = 8.8 @xmath27 5.3 kpc ) is indistinguishable from that of the whole sample .
in a hierarchical scenario of galaxy formation where one would expect that the size of galaxies evolve with the cosmic time as was reported in a number of detailed observations of galaxies at different redshift ( e.g. @xcite ) the size of the galaxies ( as derived from the half light radius ) ranges from compact ( few kpc ) objects up to extended galaxies ( 10 - 15 kpc ) ; in the observed redshift range we do not find any significant trend of change of the galaxy size with z. in some cases we found a significantly larger than average galaxy radius ( see fig . [
fig : hlr ] ) that is likely due to the presence of an extended halo .
these , and other peculiarities will be investigated in @xcite .
an interesting controversial issue in the study of qso and galaxies is the relationship between the nucleus and host galaxy luminosity . assuming that quasars emit in a relatively narrow range of eddington ratio and that the bh mass is correlated with the mass of the galaxy one would expect to find a correlation between nucleus and host galaxy luminosity . in figure
[ fig : mhn ] we show the comparison between the absolute magnitude of the nucleus and that of the host galaxy as derived from our image analysis for resolved and marginally resolved objects .
both luminosities were k - corrected and refer to rest frame r band . with our qso sample we can explore a range of nuclear luminosity between m(r ) @xmath1 -20 and m(r ) @xmath1 -24 ( average @xmath0m(r)@xmath39 @xmath2 = -22.58 @xmath27 0.80 ) . in this luminosity range
there is not a significant correlation between the two quantities ( see fig .
[ fig : mhn ] ) .
the same result is derived if we include the compilation of low redshift quasars observed by hst @xcite that extends to higher qso luminosity .
the only exceptions to this behaviour appears to be for few high luminosity ( m(r ) @xmath1 26 ) quasars that are hosted in very high luminosity galaxies .
a similar behaviour was also noted by @xcite from the comparison of a collection of seyfert galaxies and low z qso data and interpreted as a luminosity / host - mass limit .
if applied to our sample this suggests that a limit is reached when the nucleus emits ( in the r band ) a power corresponding to a factor 3 - 5 higher than the luminosity of the whole host galaxy .
the same behaviour was also observed , albeit in a smaller sample , for high redshift quasars @xcite and confirms that an intrinsic range of accretion together with different mechanisms for low power emission may concur to destroy the correlation .
moreover it is worth to note that if the bh mass is related only with the bulge mass / luminosity then one would expect additional disruption of the above correlation between bh and galaxy masses ( see also discussion in the next section ) .
a long debated question concerning the properties of the galaxies hosting quasar is its morphology ( see e.g. @xcite and references therein ) .
do quasars inhabit both disc and bulge dominated galaxies ?
this question was debated for long time since the poor spatial resolution of the observations combined with the bright nuclei hindered the clear nature of the qso hosts .
the original idea that considered radio loud qso being hosted by ellipticals while radio quiet quasars hosted in spiral galaxies is clearly not consistent with the observations that show a more complex scenario . in the era of hst images
it was clear that at low redshift qso are found in both types of galaxies spiral and ellipticals and also in complex morphology and interacting galaxies @xcite .
it was also suggested that there may be a relationship between qso luminosity and host galaxy morphology such that all the radio - loud quasars , and all the radio - quiet quasars with nuclear luminosities m@xmath40 -24 , are massive bulge - dominated galaxies , @xcite . from the analysis of our large dataset we found that 309 out of 416 targets are well resolved ( see above discussion ) , however , to be able to constrain the morphology of the host galaxy it is needed that the flux from the surrounding nebulosity is well detected up to large radii ( faint surface brightness ) where the two models ( @xmath41 and exponential disk ) differ significantly . in order to classify the morphology of the host galaxies we can use the values of sersic index obtained from the best fit . in addition
we also performed a fit of all objects assuming the host galaxy is a pure elliptical or a pure disk and then compared the @xmath30 of the two fit .
this kind of analysis can yield only a preliminary indication of the morphology of the host galaxies since in general both spheroidal and disk components may be present . in order to better characterise the host galaxies we performed a detailed visual inspection of all resolved targets using the whole information available : images , contour plots , fit of the brightness profile , ellipticity .
100 qso host galaxies from hst observations @xcite ( filled green triangles : inverted triangles for radio loud objects ) diagonal blue lines represent the loci of fixed ratio between the nucleus and host galaxy luminosity ( in the r band ) at constant levels of 1 ( solid line ) , 2.5 ( dashed line ) , 6.25 ( dotted line ) . ] the morphological classification of galaxies is an important indicator of many physical processes in galaxies and the galaxy zoo project is a clear example @xcite although it is somewhat a subjective process .
for instance note that the various tools for automatic classification or the galaxy zoo web based one can only give a rough estimate of the morphology ( see e.g. @xcite for full discussion ) .
for this reason we used the recipe of @xcite to classify our host galaxies in the same classification scheme of the carnegie atlas of galaxies @xcite and of rc3 ( @xcite ) . due to
the presence of the central nuclear source we restricted our t types to five main classes t=-5,-4 for all early type galaxies , t=-1 for e / s0 t=0 for s0 galaxies and t=1,2 for late type galaxies .
our classifications take into account both the visual inspection of the i band image and the luminosity profile . from this analysis
we find that the morphology of the host galaxies is rather complex with both disk and spheroidal components often present in these galaxies .
of the 314 resolved targets about 113 objects ( 37% ) are dominated by the bulge component , 129 objects ( 42% ) have a conspicuous disk structure , 64 objects ( 21% ) exhibit a mixed ( bulge plus disk ) features . for @xmath1100 objects ( 32% ) a number of complex features ( lens , tidal distortion , bars , close companions ) are also present .
detailed analysis of these morphological structures will be presented in another paper of this series @xcite .
this morphological classification of the host galaxies is summarised in table [ tab : abs ] .
massive black holes ( bhs ) are ubiquitously found in the centre of massive galaxies and their masses show correlations with large - scale properties of the host galaxies , namely , the stellar velocity dispersion , the luminosity , and the mass of the spheroidal component @xcite .
these relations have been interpreted as the outcome of a joint evolution between bhs and their host galaxies and are therefore potentially of great importance for the understanding of the processes that link nuclear activity to galaxy formation and evolution @xcite .
our large and homogeneous dataset allows us to investigate this relationship for low redshift quasars . for the bh mass we adopted the measurements obtained by @xcite who estimate the virial bh mass using the fwhm of @xmath42 and continuum luminosity ( @xcite ) for all the qso in sdss dr7 with @xmath43 ( @xcite ) .
all the spectra in our sample were visually inspected and 31 objects ( @xmath17% ) have been removed from the sample because of very low s / n ratio of the spectra . for three cases ( i.e. objects 311 , 349 and 365 )
we have done a new measurement of the bh mass . since these bh masses are derived from single - epoch virial mass estimates and assume an indirect measurement of the size of the blr from its relationship with the continuum luminosity the individual values may have large uncertainty . for our qso sample the quoted errors of bh mass by @xcite taking into accounts various effects ( see @xcite for details ) range on average from 0.1 dex up to about 0.4 dex ( mean error 0.17 dex ) with even larger errors in few cases . note that this uncertainty includes neither the statistical uncertainty ( @xmath440.3 - 0.4dex ) from virial mass calibrations , nor the systematic uncertainties with these virial bh masses . in figure
[ fig : mbhmh1 ] we report the relationship between the black hole mass ( m@xmath45 ) and the absolute magnitude of the host galaxy in the r band for all resolved quasars that have good s / n spectra ( see above ) .
the absolute magnitude m(r ) of the host galaxies are in the range -22 to -24 and bh masses between 10@xmath46 to 10@xmath47 m@xmath48 .
we search for possible evolution with the redshift of the m(r ) - m(bh ) relation and report in figure [ fig : mbhmhz ] the comparison of the relation for different redshift intervals . from our dataset
we do not find any significant evolution of the m(bh ) - m(r ) relation from z @xmath1 0.2 to z @xmath1 0.5 . on average the bh masses
are found systematically lower than the value expected for the host galaxy luminosity with respect to the m(r ) - m(bh ) relation established for local ( inactive ) galaxies .
in addition there is a large scatter of bh masses at the same galaxy luminosity .
for the galaxies of absolute magnitude -22 @xmath0 m(r ) @xmath0 -24 the bh mass is spread over about 2 order of magnitudes .
we argue that this is an indication that the bh mass is not well correlated with the total mass of the galaxy .
if the bh mass is linked only with the spheroidal component then the correlation would be significantly improved .
indeed we have many quasars that are hosted in galaxies with a significant disk component therefore their bulge ( or spheroid component ) represent only a fraction of the total luminosity of the galaxy . in figure
[ fig : mbhmh2 ] we show the m(bh )
m(r ) ( estimated bulge ) relation taking into account the effects of disk / bulge components .
for each resolved quasar we have classified the morphological type of the host galaxy as described in section 4 .
we have then evaluated the fraction of the bulge to total galaxy luminosity in a range between 1 to 0.3 following the above morphological classification . in figure
[ fig : mbhmh2 ] we show the relationship between m(bh ) and the estimated bulge luminosity that exhibits a significant correlation .
this indicates that the bh mass is linked only with the bulges mass / luminosity and not ( or only modestly ) with the total mass of the galaxy .
this result is also supported by the comparison of the m@xmath45 host galaxy relation for 25 low redshift ( z @xmath0 0.2 ) quasars @xcite for which the bh mass was derived from reverberation mapping technique and the host galaxy properties were obtained from acs hst images . for about half of these objects the quasar host galaxy exhibits a significant disc component and a bulge to total galaxy ratio was derived @xcite . for the rest of objects a pure bulge ( elliptical ) component was derived .
it turns out that these data well overlap with our relationships for m(bh ) - host galaxy ( see figure [ fig : mbhmh1 ] ) and m(bh ) - bulge ( see figure [ fig : mbhmh2 ] ) it is worth to note that at the lowest bh masses we find qso with relatively luminous host galaxies ( see figure [ fig : mbhmh2 ] ) while those observed by bentz et al .
, at similar bh masses , are significantly less luminous and lie close to the local m(r ) - m(bh ) relation
. since our qso with small bh masses are well resolved it is unlikely that their host luminosity be overestimated . on the other hand note that the bh masses in bentz et al .
are measured by reverberation mapping that resolves the influence of the black hole in the time domain through spectroscopic monitoring of the continuum flux variability and the delayed response , in the broad emission - line flux .
we thus argue that some bias could be in place using the virial method to derive m(bh ) for low luminosity ( likely lowest bh masses ) qso . 1 from decarli et al . 2010a .
the blue pentagons represent the sample of quasars observed with hst @xcite and for which the bh mass was derived from reverberation mapping technique @xcite . ]
but for the estimated host galaxy bulge component . ]
we have investigated the properties of the host galaxies from a large ( @xmath1 400 objects ) and homogeneous dataset of low redshift ( z @xmath0 0.5 ) quasars using th sdss images in the stripe82 region that are significantly deeper that standard sdss data .
the 2d analysis of the images allowed us to well resolve the quasar host for 3/4 of the objects in the sample , marginally resolve other 40 quasars and derive limits for the galaxy luminosity for the unresolved targets ( 60 objects ) .
the following properties of quasar hosts are derived : 1 .
the luminosity of the host galaxies of low z quasars span a range from m(r ) @xmath1 -21.5 to m(r ) @xmath1 -24.0 ; the bulk of the host galaxies are located in the region corresponding to m*-1 and m*-2 ; there is a mild increase of the host luminosity with the redshift that is consistent with the passive evolution of the underlying stellar population 2 .
the morphology of the host galaxies turned out to be rather complex with both bulge and disc dominated galaxies ; about one third of the objects in our sample show features characteristics of bulge and disc components 3 .
irrespective of the host morphology the size of the galaxies ( as derived from the half light radius ) ranges from compact ( few kpc ) objects up to extended galaxies ( 10 - 15 kpc ) ; in the observed redshift range we do not find any significant trend of change of the galaxy size with z 4 .
the nuclear and host galaxy luminosities are not correlated suggesting that accretion rate , bh mass , and galaxy masses and morphology combine together to smear significantly the correlation between bh and host masses 5 . the bh mass of quasars estimated from the qso continuum luminosity and the width of the broad emission lines is poorly correlated with the total luminosity / mass of the whole host galaxy ; on the contrary when the fraction of bulge to disc component is considered we find a significant correlation between the bh mass and the bulge luminosity of the host .
another important source of information to characterise the properties of low redshift qso come from the analysis of their galaxy environments as compared with those of similar galaxies with no active nuclei .
these aspects will be pursued in forthcoming papers of this series @xcite .
in order to test the reliability of the image decomposition we have performed a number of mock simulations of quasars and then analysed them with the same method used for the stripe82 images . to perform the simulation of the quasars we used the advanced exposure time calculator tool .
the parameters of the sloan telescope and the global efficiency of the instrument were adopted from sdss web site .
we used read out noise of 9.5e , gain of 3.8 @xcite , exposure time of 1 hour ( subdivided in 60 exposures of 1 minute ) , sky brightness as average value of kitt peak .
statistical noise was added to the mock objects and background .
the final images are background subtracted as are stripe82 images .
three set of simulations were performed assuming different galaxy model with sersic index n=1 , 2.5 and 4 to represent disk , intermediate and elliptical host galaxies , respectively . for each dataset
we used a number of psf extracted from the psf of our images and corresponding to seeing in the range 1.0 1.4 arcsec ( fwhm ) .
then we construct the qso images as the superposition of a nucleus and a galaxy with a range of values that map the observed values .
we explored a range of nucleus / host galaxy flux ratio from 0.1 to 2 .
moreover for each galaxy model we assume effective radius of 1.0 , 1.5 , 2.0 and 2.5 arcsec ( again similar to the observed parameters ) . in figure [
fig : mock_comb ] we show a representative example for the three adopted host galaxy models of the mock simulations together with the fit obtained following the same procedure used for the real quasars . in figures [ fig : mock_n4 ] , [ fig : mock_n25 ] , [ fig : mock_n1 ] we show the comparison between the measure and the true parameters of the host galaxies for the various mock simulations .
it turns out that the magnitude of the host galaxy is very well measured ( @xmath0 0.1 mag ) for nucleus / galaxy flux ratio smaller than 1 and seeing better than 1.2 arcsec .
also the effective radius of the host galaxy is recovered within an accuracy of 20% . for nucleus / galaxy flux ratio greater than 1 and seeing worst than 1.2 arcsec
the uncertainty is larger but still adequate ( @xmath49 ) for the results presented in this work .
but for host galaxy with sersic index n = 2.5 . , title="fig : " ] but for host galaxy with sersic index n = 2.5 .
, title="fig : " ] but for host galaxy with sersic index n = 1 .
, title="fig : " ] but for host galaxy with sersic index n = 1 .
, title="fig : " ] [ cols= " > , < , > , > , > , > , < , > , > , > " , ] note : m@xmath50 nucleus and host magnitudes are in the r band k - corrected .
+ r@xmath51 is effective radius in kpc ; c is the e class type : r = resolved object , u = unresolved , m = marginally resolved , x = discarded object + e is the ellipticity of the host galaxy and morph type is the following : e = elliptical dominant , d = disk dominant , n = not classifiable .
+ [ tab : abs ] [ fig : fit2 ]
we are grateful to the anonymous referee for detailed comments that led to improve the presentation of these results .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , university of cambridge , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
abazajian , k. n. , adelman - mccarthy , j. k. , ageros , m. a. , et al .
2009 , apj , 182 , 543 annis , j. , soares - santos , m. , strauss , m. a. , et al .
2011 , arxiv:1111.6619 bahcall j.n .
, kirhakos s. , saxe d.h . ,
schneider d.p . , 1997 , apj , 479 , 642 becker , r. h. , helfand , d. j. , white , r. l. , gregg , m. d. , & laurent - muehleisen , s. a. 2012 , vizier online data catalog , 8090 , 0 becker , r. h. , gregg , m. d. , hook , i. m. , et al .
1997 , apjl , 479 , l93 bernardi m. , sheth r.k .
, tundo e. , hyde j.b . , 2007 , apj,660 , 267 bettoni d. , falomo r. , fasano g. , govoni f. , 2003 , a&a , 399 , 869 bettoni , d. falomo , r. , karhunen , k. , kotilainen j.k . , uslenghi , m. 2013 , in preparartion .
bentz , m. c. , peterson , b. m. , netzer , h. , pogge , r. w. , & vestergaard , m. 2009 , apj , 697 , 160 bentz , m. c. , peterson , b. m. , pogge , r. w. , & vestergaard , m. 2009 , apjl , 694 , l166 bhm a. , et al .
, 2013 , a&a , 549 , a46 bouwens r. , silk j. , 2002 , apj , 568 , 522 bouwens r. j. , illingworth g. d. , blakeslee j. p. , broadhurst t. j. , franx m. , 2004 , apj , 611 , l1 cales , s. l. , brotherton , m. s. , shang , z. , et al .
2011 , apj , 741 , 106 cisternas , m . , jahnke , k . ,
bongiorno , a .
, et al . , 2011 , apj , 741 , l11 decarli r. , falomo r. , kotilainen j. k. , hyvnen t. , uslenghi m. , treves a. , 2012 , adast , 2012 , decarli r. , falomo r. , treves a. , labita m. , kotilainen j. k. , scarpa r. , 2010 , mnras , 402 , 2453 decarli r. , falomo r. , treves a. , kotilainen j. k. , labita m. , scarpa r. , 2010 , mnras , 402 , 2441 de vaucouleurs g. , de vaucouleurs a. , corwin h. g. , jr .
, buta r. j. , paturel g. , fouqu p. , 1991
, rc3 , springer - verlag : new york dunlop j.s . ,
mclure r.j . ,
kukula m.j . , et al . , 2003 , mnras , 340 , 1095 falomo r. , kotilainen j.k . , pagani c. , scarpa r. , treves , a. , 2004 , apj , 604 , 495 falomo r. , treves a. , kotilainen j.k . , scarpa r. , uslenghi m. , 2008 , apj , 673 , 694 francis p.j .
, hewett p.c . , foltz c.b .
, et al . , 1991 , apj , 373 , 465 ferrarese , l . , 2006 , joint evolution of black holes and galaxies , eds .
m. colpi et al .
( taylor & francis group ) , p. 1
aj , 135 , 338 floyd d.j.e .
, kukula m.j . , dunlop j.s . , et al . , 2004 , mnras , 355 , 196 floyd d.j.e . , dunlop j.s
, kukula m.j . , et al . , 2013 , mnras , 429 , 2 gunn j. e. , et al . , 1998 , aj , 116 , 3040 letawe , y. , letawe , y. and magain , p. 2010
, mnras , 403 , 2088 lintott , c. , schawinski , k. , bamford , s. , et al . 2011 ,
mnras , 410,166 jahnke k. , et al . , 2009 ,
apj , 706 , l215 karhunen , k. kotilainen , falomo , r. , bettoni , d .
usleghi , m. 2013 , in preparation kotilainen j.k . ,
falomo r.,2004 , a&a , 424 , 107 kotilainen j.k .
, falomo r. , labita m. , treves a. , & uslenghi m. , 2007 , apj 660 1039 kotilainen j.k .
, falomo , r. , decarli , r. , treves , uslenghi , m and scarpa , r. , apj 703 1663 kotilainen j.k .
, falomo , r. , bettoni , d. , karhunen , k. , uslenghi , m. 2013 arxiv:1302.1366 .
kukula m.j .
, dunlop j.s . , mclure r.j . , et al . , 2001 ,
mnras , 326 , 1533 marconi a. , hunt l.k . , 2003 , apj letters , 589 , 21 mannucci , f. , basile , f. , poggianti , b.m . ,
2001 , mnras , 326 , 745 matsuoka y. , strauss m. a. , price t. n. , iii , didonato m. s. , 2014 , apj , 780 , 162 merloni , a .
, bongiorno , a . ,
bolzonella , m .
et al . , 2010 , apj 708 , 137 mcleod k. k. , mcleod b. a. , 2001 , apj , 546 , 782 nair , p. b. , & abraham , r. g. 2010 , apjl , 714 , l260 nakamura , o. , fukugita , m. , yasuda , n. , et al .
2003 , aj , 125 , 1682 ono y. , et al . , 2012 ,
arxiv , arxiv:1212.3869 peng , c.y . ,
impey , c.d . ,
rix , h .- w . , et al .
2006 , apj , 649 , 616 ridgway s. , heckman t. , calzetti d. , lehnert m. , 2001 , apj , 550 , 122 sandage , a. , & bedke , j. 1994 , the carnegie atlas of galaxies .
volumes i , ii .
, by sandage , a. ; bedke , j .. carnegie institution of washington publ .
, no . 638 shen , y. , richards , g.t . , strauss , m.a .
2011 , apjss , 194 , 45 serber , w. , bahcall , n. , mnard , b. , & richards , g. 2006 , apj , 643 , 68 schneider , d. p. , richards , g. t. , hall , p. b. , et al .
2010 , aj , 139 , 2360 smith , r. j. , boyle , b. j. , & maddox , s. j. 2000 , mnras , 313 , 252 strand , n. e. 2009 , ph.d .
thesis targett t. a. , dunlop j. s. , mclure r. j. , 2012 , mnras , 420 , 3621 trujillo i. , et al . , 2006 , apj , 650 , 18 uslenghi m. & falomo , r. 2008 , proc .
erice , in press .
vestergaard m. , peterson b. m. , 2006 , apj , 641 , 689 vestergaard m. , osmer p. s. , 2009 , apj , 699 , 800 vron - cetty m .- p . , vron p. , 2010 ,
a&a , 518 , a10 wold m. , lacy m. , lilje p. b. , serjeant s. , 2001 , mnras , 323 , 231 york , d. g. et al .
2000 , aj , 120 , 1579 zwaan m. a. , van der hulst j. m. , briggs f. h. , verheijen m. a. w. , ryan - weber e. v. , 2005 , mnras , 364 , 1467 zakamska , n. l. , strauss , m. a. , krolik , j. h. , et al .
2006 , aj , 132 , 1496 zibetti s. , mnard b. , nestor d. b. , quider a. m. , rao s. m. , turnshek d. a. , 2007 , apj , 658 , 161 | we present a photometrical and morphological study of the properties of low redshift ( z @xmath0 0.5 ) quasars based on a large and homogeneous dataset of objects derived from the sloan digital sky survey ( dr7 ) .
this study over number by a factor @xmath1 5 any other previous study of qso host galaxies at low redshift undertaken either on ground or on space surveys .
we used @xmath1 400 quasars that were imaged in the sdss stripe82 that is up to 2 mag deeper than standard sloan images .
for these quasars we undertake a study of the host galaxies and of their environments . in this paper
we report the results for the quasar hosts .
we are able to detect the host galaxy for more than 3/4 of the whole dataset and characterise the properties of their hosts .
we found that qso hosts are dominated by luminous galaxies of absolute magnitude m*-3 @xmath0 m(r ) @xmath0 m*. for the unresolved objects we computed a upper limit to the host luminosity . for each
well resolved quasar we are also able to characterise the morphology of the host galaxy that turn out to be more complex than what found in previous studies .
qso are hosted in a variety of galaxies from pure ellipticals to complex / composite morphologies that combine spheroids , disk , lens and halo .
the black hole mass of the quasar , estimated from the spectral properties of the nuclei , are poorly correlated with the total luminosity of the host galaxy . however , taking into account only the bulge component we found a significant correlation between the bh mass and the bulge luminosity of the host .
[ firstpage ] |
it is first convenient to write the spin - operator in the particle - hole - conjugate form @xmath159 .
substitution of the composite form @xmath34 for the creation operator of the correlated electron _
above _ half filling then yields the expression @xmath160 .
next , replacing the operator @xmath161 with its expectation value , @xmath162 , yields the approximation @xmath163 for the spin operator . here , @xmath127 denotes the concentration of mobile electrons per orbital .
last , we shall also neglect on - site repulsion between mobile electrons on the @xmath69 and @xmath70 orbitals , @xmath164 , which is the energy cost of a fully occupied iron site .
this approximation should be valid in the dilute limit , @xmath165 . as mentioned in the letter
, it is also convenient to next rotate the spins quantized along the @xmath41 axis by an angle @xmath53 about the @xmath54 axis on one of the antiferromagnetic sublattices in the hidden magnetic order shown by the inset to fig .
1a ; e.g. , @xmath166 and @xmath167 , for @xmath168 that lie in the down - spin sublattice .
this decouples spins between the two hidden antiferromagnetic sublattices@xcite .
let us now turn off nearest - neighbor intra - orbital hopping in the two - orbital @xmath12-@xmath13 model , eq .
( 1 ) in the letter : @xmath50 .
mean fields among the schwinger bosons are pair amplitudes across the antiferromagnetic links@xcite : @xmath55 , @xmath56 and @xmath57 . here
, the superscripts @xmath45 and @xmath46 denote intra - orbital ( @xmath47 ) and inter - orbital ( @xmath48 ) links , while the subscripts @xmath60 , @xmath61 and @xmath62 denote on - site , nearest neighbor and next - nearest neighbor links . on the other hand , mean fields among the schwinger bosons are hopping amplitudes across the ferromagnetic links@xcite : @xmath58 and @xmath59 .
last , nearest - neighbor hopping of electrons across the two orbitals is accounted for by the mean field among slave fermions @xmath63 , which has @xmath5-wave symmetry : @xmath169 .
the dynamics of free schwinger bosons is governed by the hamiltonian @xmath170 + \omega_{afm}(k ) [ b_s^{\dagger}(k ) b_s^{\dagger}(-k ) + b_s(-k ) b_s(k)]\},\ ] ] with diagonal and off - diagonal matrix elements @xmath171[1-e^{ik_0}\gamma_{1+}({\bm k } ) ] - 4 j_2^{\prime\parallel } q_2^{\parallel}[1-\gamma_2({\bm k})]\\ \omega_{afm}(k ) & = & j_0^{\prime } q_0 e^{i k_0 } + 4 j_1^{\prime\parallel } q_1^{\parallel } \gamma_{1+}({\bm k } ) + 4 j_2^{\prime\perp } q_2^{\perp}e^{ik_0}\gamma_2({\bm k } ) , \end{aligned}\ ] ] while the dynamics of free slave fermions is governed by the hamiltonian @xmath66 , with the energy eigenvalues @xmath172 from here on we set @xmath173 . above
, @xmath67 is the 3-momentum for these excitations , with corresponding destruction operators @xmath174 and @xmath175 . here
, @xmath176 denotes the number of sites - orbitals on the square lattice of @xmath177 iron atoms , while the indices @xmath60 and @xmath61 denote the @xmath69 and @xmath70 orbitals @xmath24 .
the quantum numbers @xmath68 and @xmath53 therefore represent the @xmath14 and the @xmath71 orbitals . also above ,
@xmath80 and @xmath81 , with @xmath82 . the infinite-@xmath18 constraint , eq .
( 2 ) in the letter , is enforced on _ average _ over the bulk of the system by the boson chemical potential , @xmath178 , while the chemical potential of the slave fermions , @xmath100 , sets the concentration of mobile electrons per site - orbital , @xmath127 .
last , the mean - field approximation of mobile electrons in the spin operator ( [ s_mf_spin_operator ] ) results in effective heisenberg spin - exchange coupling constants@xcite @xmath179 .
the solution to the above mean field theory is achieved by making the standard bogoliubov transformation of the boson field@xcite : @xmath180 , with @xmath181 and @xmath182 , where @xmath88 is the energy of the boson ( @xmath30 ) . enforcing the infinite-@xmath18 constraint [ eq .
( 2 ) in the letter ] on average then results in ideal bose - einstein condensation ( bec ) of the schwinger bosons into degenerate groundstates at @xmath72 and @xmath73 as temperature @xmath183 , in which case @xmath184 .
( see letter , fig .
1a , and fig .
[ s_meanfield_b ] . )
all five mean fields among the schwinger bosons take on the unique value @xmath76 at large-@xmath75 under ideal bec@xcite .
slave fermions in @xmath14 and @xmath15 orbitals condense inside of circular fermi surfaces centered at wavenumbers @xmath7 and @xmath6 , respectively , at low electron doping @xmath84 , with fermi wave vector @xmath83 .
( see the inset to fig .
1b in the letter . )
the mean inter - orbital electron hopping amplitude is then approximately @xmath85 . equation ( 3 ) in the letter for the dynamical spin correlation function @xmath185 of the hidden nel half metal is a direct application of the auerbach - arovas expression for the auto - correlation function @xmath186 under bec by the schwinger bosons@xcite , multiplied by a factor of two because of spin isotropy .
the result notably coincides with that obtained within the linear spin - wave approximation at the large-@xmath75 limit@xcite .
figure [ s_meanfield_b ] gives the hidden - order counterpart to the spectrum of true spinwaves near the qcp predicted by this mean - field approximation , fig .
1a in the letter .
also , within the above mean field theory , the one - electron propagator is given by the convolution of the propagator for slave fermions ( @xmath33 ) with the conjugate propagator for schwinger bosons ( @xmath32 ) in 3-momentum and in frequency : @xmath187 . here , the propagator for free slave fermions reads @xmath188^{-1}$ ] , while the propagator for free schwinger bosons reads @xmath189^{-1 } -({\rm sinh } \
, \theta_k)^2 [ \omega + \omega_b(k)]^{-1}$ ] . the conjugate propagator for free schwinger bosons evolving backwards in time is then @xmath190^{-1 } + ( { \rm sinh } \
, \theta_k)^2 [ \omega - \omega_b(k)]^{-1}$ ] .
after rewriting the resulting products of poles as sums / differences of poles , standard summations of matsubara frequencies yield expression ( 5 ) given in the letter . there , the identities @xmath191 and @xmath192 have been used .
hidden magnetic order of the type depicted by the inset to fig .
1a in the letter is predicted by the two - orbital heisenberg model over the square lattice in the large-@xmath75 limit for exhange coupling constants that exhibit diagonal frustration , at weak to moderate hund coupling@xcite : e.g. , @xmath42 , @xmath43 , @xmath193 , and @xmath194 , where @xmath195 is the quantum - critical hund coupling at which the spin gap associated with commensurate spin - density wave ( csdw ) order collapses to zero .
add now electrons with ideal nearest - neighbor hopping ; e.g. , @xmath50 , @xmath196 and @xmath197 .
spin - polarized electrons remain within each antiferromagnetic sublattice in such case .
the critical hund coupling is given by @xmath93 within the mean - field approximation .
adding mobile electrons thereby stabilizes the hidden nel order .
figure [ s_meanfield_b ] reveals the goldstone mode at nel wavenumber @xmath17 expected from such hidden antiferromagnetic order for the corresponding half metal state of the two - orbital @xmath12-@xmath13 model within the mean - field approximation@xcite .
it appears as a divergence in the imaginary part of the transverse susceptibility , eq .
( 3 ) in the letter , for hidden spin , @xmath95 .
recall that spin-@xmath0 moments live on the @xmath198 orbitals .
the schwinger - boson - slave - fermion mean - field approximation for the two - orbital @xmath12-@xmath13 model employed in the letter also predicts coherent electron bands that result in fermi surface pockets centered at csdw momenta @xmath199 and @xmath200 .
( see the inset to fig .
1b in the letter . ) below , we compare this mean field theory to exact results in the absence of hund s rule , @xmath201 , where the hidden half metal state is most stable .
_ hund s rule absent .
_ figure [ s_exact_32_30]a compares exact results for the low - energy spectrum of the frustrated heisenberg model on a periodic @xmath116 lattice of iron atoms with @xmath69 and @xmath70 orbitals to the spin - wave spectrum predicted by schwinger - boson mean field theory for hidden magnetic order , eq .
( 3 ) in the letter .
heisenberg exchange coupling constants are set by fig .
[ s_meanfield_b ] , but without hund s rule : @xmath201 . also , the concentration of mobile electrons per site - orbital is set to @xmath202 in all mean - field expressions .
black states in fig .
[ s_exact_32_30]a have even parity under orbital exchange @xmath119 , while red states have odd parity under it .
black spin-@xmath61 states therefore represent true spin fluctuations , while red spin-@xmath61 states represent hidden spin fluctuations .
notice that the predicted spin - wave spectrum for the hidden nel state traced by the dashed lines in fig .
[ s_exact_32_30]a successfully describes the dispersion of the exact spin-@xmath61 states at low energy states there in figs .
[ s_exact_32_30]a , b . ] .
notice also the tower in fig .
[ s_exact_32_30]a beginning with the spin-@xmath60 groundstate at zero 2d momentum , the spin-@xmath61 first - excited state at 2d momentum @xmath203 , the spin-@xmath62 second - excited state back at zero 2d momentum , and the spin-@xmath204 excited state back at 2d momentum @xmath203 .
this tower of spin-@xmath205 states clearly coincides with multiply - occupied states of the hidden order spinwave , which is occupied @xmath205 times .
figure [ s_exact_31 ] shows the exact low - energy spectrum of one electron more than half filling governed by the two - orbital @xmath12-@xmath13 model , eq .
( 1 ) in the letter , in the absence of hund s rule .
heisenberg exchange coupling constants coincide with those in figs .
[ s_meanfield_b ] and [ s_exact_32_30 ] , while hopping matrix elements are set to @xmath117 , @xmath206 and @xmath207 .
red states are even under orbital exchange @xmath119 , while blue states are odd under it .
the solid blue line depicts the @xmath15 half metal band predicted by schwinger - boson - slave - fermion mean field theory at the limit towards half filling , @xmath208 , but with ideal electron hopping , @xmath50 .
the dashed lines trace the dispersion of emergent hole excitations predicted by eq .
( 6 ) and fig . 1b of the letter .
they successfully describe the dispersion of the exact spin-@xmath0 groundstates per orbital quantum number in the absence of hund s rule .
notice , however , the first - excited states per momentum that carry spin @xmath209 in fig .
[ s_exact_31 ] .
the pairs of spin-@xmath0 and spin-@xmath209 states that they make up per momentum can be understood as the result of the addition of angular momentum between a spin-@xmath0 electron at csdw wavenumbers and a spin-@xmath61 spinwave in the half metal@xcite .
last , fig .
[ s_exact_32_30]b compares the exact spectrum of two electrons more than half - filling in the absence of hund s rule with the spin - excitation spectrum predicted by schwinger - boson - slave - fermion mean field theory at ideal electron hopping @xmath50 .
notice that the tower of @xmath210 and @xmath204-occupied hidden - order spinwave states persists@xmath211 .
comparison with fig .
[ s_exact_32_30]a indicates that a gap separates out the tower of lowest - energy states in the case of two mobile electrons .
schwinger - boson mean field theory predicts a divergent spectral weight for hidden spinwaves ( `` red '' ) at wavenumber @xmath17 and a relatively weak spectral weight of order unity for them at zero 2d momentum .
inspection of fig .
[ s_exact_32_30]b indicates that the gap also separates spin-@xmath61 excitations expected to have large spectral weight according to mean field theory from spin-@xmath61 excitations expected to have weak spectral weight .
we therefore conclude that schwinger - boson - slave - fermion mean field theory is a valid approximation for the two - orbital @xmath12-@xmath13 model in the case of the hidden half metal state depicted by the insets to fig . 1 of the letter .
( cf . refs .
@xcite and @xcite . )
in particular , in the absence of hund s rule , fig . [ s_exact_32_30 ] demonstrates that it works well for spin-@xmath61 states at both half filling and in the case of two mobile electrons .
in the absence of hund s rule , fig .
[ s_exact_31 ] demonstrates that schwinger - boson - slave - fermion mean field theory also works well in the case of one mobile hole for spin-@xmath0 states . _
. _ figure [ s_exact_32_qcp ] shows the exact spectrum of the same two - orbital heisenberg model that corresponds to fig .
[ s_exact_32_30]a , but at the putative quantum - critical point . here
, the hund coupling is tuned to the critical value @xmath212 at which the lowest energy spin-@xmath61 states at csdw wavenumbers @xmath6 and @xmath7 become degenerate with the lowest - energy spin-@xmath61 state at nel wavenumber @xmath17 .
the former states are true spin fluctuations , with even parity under @xmath119 , while the latter state is a hidden spin fluctuation , with odd parity under @xmath119 . | we model a single layer of heavily electron - doped fese by spin-@xmath0 moments over a square lattice of iron atoms that include the @xmath1 and @xmath2 orbitals , at strong on - site coulomb repulsion . above half filling , we find emergent hole bands below the fermi level at the brillouin zone center in a half metal state characterized by hidden magnetic order and by electron - type fermi surface pockets at wavenumbers that double the unit cell along the principal axes .
exact calculations with two mobile electrons find evidence for isotropic cooper pairs that alternate in sign between the electron bands and the emergent hole bands .
_ introduction . _
the discovery of superconductivity in iron - pnictide materials has uncovered a new path in the search for high - temperature superconductors@xcite .
superconductivity has been observed recently in a single layer of fese on a doped srtio@xmath3 ( sto ) substrate@xcite below critical temperatures as high as @xmath4 k @xcite .
electronic conduction originates from the 3@xmath5 orbitals of the iron atoms , which form a square lattice .
angle - resolved photo - emission spectroscopy(arpes ) , in particular , reveals circular electron - type fermi surface pockets centered at wave numbers @xmath6 and @xmath7 that lie along the principal axes of the iron lattice , where @xmath8 is the lattice constant@xcite . unlike the case of most iron - pnictide materials , however , arpes also finds that hole bands centered at zero two - dimensional ( 2d ) momentum lie well below the fermi level in the case of single - layer fese / sto . at low temperature
, it also finds an isotropic gap at the electron fermi surface pockets@xcite , which is confirmed by scanning tunneling microscopy ( stm)@xcite . the same set of phenomena have been recently observed below critical temperatures in the range @xmath9-@xmath10 k at the surfaces of intercalated fese@xcite , of alkali - metal dosed fese@xcite , and of voltage - gate tuned thin films of fese@xcite .
comparison with bulk fese , which has a much lower critical temperature of @xmath11 k , strongly suggests that the high - temperature superconductivity exhibited above is due to a new 2d groundstate that appears after heavy electron doping .
calculations based on the independent - electron approximation fail to describe the fermi surfaces in single - layer fese / sto .
in particular , density - functional theory ( dft ) typically predicts that the hole bands centered at zero 2d momentum cross the fermi level@xcite .
dft also fails to account for a nearby mott insulator phase at low electron doping in voltage - gate tuned thin films of fese and in single - layer fese / sto@xcite .
the previous suggests that the limit of strong electron - electron interactions is a better starting point to describe superconductivity in heavily electron - doped fese .
below , we propose that the hole bands observed by arpes below the fermi level at the brillouin zone center in a surface layer of fese are examples of emergent phenomena .
the latter is revealed by both mean - field and exact calculations of the one - electron spectrum in a two - orbital @xmath12-@xmath13 model that includes only degenerate @xmath14 and @xmath15 electron bands centered at wavenumbers @xmath7 and @xmath6 , respectively .
local spin-@xmath0 moments live on @xmath16 orbitals , on the other hand , which yields isotropic magnetism .
emergent hole bands approach the fermi level at zero 2d momentum as hund coupling increases inside of a half metal phase that is characterized by hidden nel order per @xmath16 orbital and by electron - type fermi surface pockets inherited from the above band structure ( fig .
[ meanfield ] ) .
emergent hole bands at wavenumber @xmath17 are also predicted , but they lie below the former ones in energy .
last , exact calculations of two mobile electrons in the two - orbital @xmath12-@xmath13 model find evidence for isotropic cooper pairs on both the electron pockets and on the emergent hole bands below the fermi level as hund coupling approaches a quantum critical point ( qcp ) at which commensurate spin - density wave ( csdw ) nesting begins .
the sign of the cooper pair wavefunction notably alternates between the electron and hole bands@xcite . _ local moment model . _
our starting point is a two - orbital @xmath12-@xmath13 model over the square lattice , where the on - site - orbital energy cost @xmath18 tends to infinity@xcite : @xmath19 + \sum_{\langle\langle i , j \rangle\rangle } j_2^{\alpha,\beta } { \bm s}_{i , \alpha } \cdot { \bm s}_{j , \beta } \nonumber \\ & + \sum_i ( j_0 { \bm s}_{i , d-}\cdot { \bm s}_{i , d+ } + u_0^{\prime } { \bar n}_{i , d+ } { \bar n}_{i , d- } \bigr ) . % \\ \label{tj}\end{aligned}\ ] ] above , @xmath20 is the spin operator that acts on spin @xmath21 states of @xmath22 and @xmath23 orbitals @xmath24 in iron atoms at sites @xmath25 .
repeated orbital and spin indices in the hopping and heisenberg exchange terms above are summed over .
nearest neighbor and next - nearest neighbor heisenberg exchange across the links @xmath26 and @xmath27 is controlled by exchange coupling constants @xmath28 and @xmath29 , respectively .
hopping of an electron in orbital @xmath24 to a nearest - neighbor orbital @xmath30 is controlled by the matrix element @xmath31 .
we adopt the schwinger - boson ( @xmath32 ) slave - fermion ( @xmath33 ) representation for the creation operator of the correlated electron at or _ above _
half filling@xcite : @xmath34 with the constraint @xmath35 enforced at each site - orbital to impose the @xmath36 limit on electrons with spin @xmath21 . finally , @xmath37 is a ferromagnetic exchange coupling constant that imposes hund s rule , while the last term in ( [ tj ] ) represents the energy cost of a fully occupied iron atom . here
@xmath38 counts singlet pairs at site - orbitals .
last , notice that @xmath39 is equivalent to a rotation of the orbitals by an angle @xmath40 about the @xmath41 axis .
spin and occupation operators remain invariant under it .
magnetism described by the two - orbital @xmath12-@xmath13 model ( [ tj ] ) is hence isotropic .
semi - classical calculations of the heisenberg model that corresponds to ( [ tj ] ) at half filling find a qcp that separates a csdw at strong hund coupling from a hidden antiferromagnet at weak hund coupling when diagonal frustration is present@xcite : e.g. @xmath42 , @xmath43 , and @xmath44 .
here , @xmath45 and @xmath46 represent intra - orbital ( @xmath47 ) and inter - orbital ( @xmath48 ) superscripts .
the hidden - order magnet shows nel spin order per @xmath49 orbital following the inset to fig .
[ meanfield]a .
ideal hopping of electrons within an antiferromagnetic sublattice , @xmath50 and @xmath51 , leaves such hidden magnetic order intact in the semi - classical limit , @xmath52 .
below , we employ a mean - field approximation of ( [ tj ] ) and ( [ constraint ] ) to study this state near the qcp .
it reveals a half metal with circular fermi surface pockets at wavenumbers @xmath6 and @xmath7 , for electrons in the @xmath15 orbital and @xmath14 orbitals , respectively .
_ spin - fluctuations , one - electron spectrum . _ following arovas and auerbach@xcite , we first rotate the spins quantized along the @xmath41 axis on _ one _ of the antiferromagnetic sublattices shown in the inset to fig .
[ meanfield]a by an angle @xmath53 about the @xmath54 axis .
this decouples the up and down spins between the two sublattices@xcite .
we next define mean fields that are set by the pattern of antiferromagnetic versus ferromagnetic pairs of neighboring spins@xcite in the hidden magnetic order : @xmath55 , @xmath56 and @xmath57 on the antiferromagnetic links versus @xmath58 and @xmath59 on the ferromagnetic links of the hidden nel state .
subscripts @xmath60 , @xmath61 and @xmath62 represent on - site , nearest neighbor and next - nearest neighbor links .
we add to that list the mean field @xmath63 for nearest - neighbor hopping of electrons across the two orbitals .
it has @xmath5-wave symmetry .
the corresponding mean - field approximation for the @xmath12-@xmath13 model hamiltonian ( [ tj ] ) then has the form @xmath64 , where @xmath65 + \omega_{afm}(k ) [ b_s^{\dagger}(k ) b_s^{\dagger}(-k ) + b_s(-k ) b_s(k)]\}\ ] ] is the hamiltonian for free schwinger bosons , and where @xmath66 is the hamiltonian for free slave fermions . here
, @xmath67 is the 3-momentum for these excitations , where the quantum numbers @xmath68 and @xmath53 represent even and odd superpositions of the @xmath69 and @xmath70 orbitals : @xmath14 and @xmath71 . enforcing the infinite-@xmath18 constraint ( [ constraint ] ) on average over the bulk
then results in ideal bose - einstein condensation ( bec ) of the schwinger bosons into degenerate groundstates at @xmath72 and @xmath73 in the zero - temperature limit : @xmath74 at large @xmath75 .
( see fig .
[ meanfield]a and supplemental material , fig . s1 . ) in such case , all five mean fields among the schwinger bosons therefore take on the unique value @xmath76 .
this results in diagonal and off - diagonal hamiltonian matrix elements @xmath77 - 4 j_2^{\parallel } [ 1-\gamma_2({\bm k})])\\ \omega_{afm}(k ) & = & ( 1-x)^2 s_0 [ j_0 e^{i k_0 } + 4 j_1^{\parallel } \gamma_{1+}({\bm k } ) + 4 j_2^{\perp } e^{ik_0}\gamma_2({\bm k } ) ] , \end{aligned}\ ] ] for free schwinger bosons , and the energy eigenvalues @xmath78 for free slave fermions . above ,
@xmath79 , while @xmath80 and @xmath81 , with @xmath82 .
slave fermions in @xmath14 and @xmath15 orbitals lie within circular fermi surfaces centered at wavenumbers @xmath7 and @xmath6 , respectively , with fermi wave vector @xmath83 at low electron doping per iron orbital , @xmath84 .
( see the inset to fig .
[ meanfield]b . )
the mean inter - orbital electron hopping amplitude is then approximately @xmath85 . the dynamical spin correlation function @xmath86 is obtained directly from the above schwinger - boson - slave - fermion mean field theory .
it is given by an auerbach - arovas expression at non - zero temperature that is easily evaluated in the zero - temperature limit @xcite , where ideal bec of the schwinger bosons into the degenerate groundstates at 3-momenta @xmath72 and @xmath73 occurs .
it is one half the transverse spin correlator , which under ideal bec and at large @xmath75 reads @xmath87^{-1 } + [ \omega_b(k ) + \omega ] ^{-1 } ) .
\label{chi_perp}\ ] ] here , @xmath88 is the energy dispersion of the schwinger bosons , and @xmath89 .
figure [ meanfield]a depicts the imaginary part of the transverse susceptibility ( [ chi_perp ] ) in the true spin channel , @xmath68 , at sub - critical hund coupling .
it reveals a spin gap at csdw wave numbers @xmath90 and @xmath91 of the form @xmath92 here , @xmath93 is the critical hund coupling at which @xmath94 . notice that inter - orbital hopping stabilizes the hidden half metal state .
the autocorrelator of the hidden spin @xmath95 , ( [ chi_perp ] ) at @xmath96 , also shows the above spin gap ( [ delta_csdw ] ) at csdw momenta , in addition to a hidden - order goldstone mode at nel wavenumber @xmath17@xcite .
the electronic structure of the hidden half metal state can also be obtained directly from the above schwinger - boson - slave - fermion mean field theory .
in particular , the one - electron propagator is given by the convolution of the conjugate propagator for schwinger bosons with the propagator for slave fermions in 3-momentum and in frequency .
a summation of matsubara frequencies yields the expression@xcite @xmath97 + n_f[\varepsilon_f ( q)-\mu]}\over{\omega + \omega_b(q - k ) - \varepsilon_f(q)+\mu } } \nonumber \\ & + \bigl({1\over 2 } { \omega_{fm}\over{\omega_b}}\bigl|_{q - k } - { 1\over 2}\bigr ) { { n_b[\omega_b ( q - k ) ] + n_f[\mu-\varepsilon_f(q)]}\over{\omega - \omega_b(q - k ) - \varepsilon_f(q)+\mu}}\bigr ] .
% \\ \label{g}\end{aligned}\ ] ] above , @xmath98 and @xmath99 denote the bose - einstein and the fermi - dirac distributions , and @xmath100 denotes the chemical potential of the slave fermions .
ideal bec of the schwinger bosons at 3-momenta @xmath101 and @xmath73 results in the following coherent contribution to the electronic spectral function at zero temperature and at large @xmath75 : @xmath102 $ ] .
it reveals degenerate electron bands for @xmath14 and @xmath15 orbitals centered at csdw wave numbers @xmath103 and @xmath104 , respectively .
the electron fermi surface pockets at @xmath105 are depicted by the inset to fig .
[ meanfield]b . at energies below the fermi level , @xmath106 ,
the remaining contribution is exclusively due to the first fermion term in ( [ g ] ) .
inspection of fig .
[ meanfield]b ( solid lines ) yields the following expression for it in the limit near half - filling , @xmath107 , at large @xmath108 @xcite : @xmath109 \delta[\omega + \epsilon_f + \omega_b(q_0-k_0,{\bm q}_{q_0 } - { \bm k } ) ] .
\label{emergent_holes}\ ] ] figure [ meanfield]b displays the emergent hole bands predicted above .
they lie @xmath110 below the fermi level , with degenerate maxima at @xmath111 and @xmath17 . here , @xmath112 is the fermi energy .
the emergent hole bands also show intrinsic broadening in frequency at zero temperature , which makes them incoherent .
outside the critical region , at large @xmath108 , the broadening is @xmath113 it remains small at the previous maxima@xcite .
last , the emergent hole bands predicted by ( [ emergent_holes ] ) are anisotropic : e.g. , the @xmath15 hole band at zero 2d momentum has mass anisotropy @xmath114 .
( cf .
ref .
@xcite . ) adding intra - orbital electron hopping , @xmath115 , brings the emergent hole bands at wavenumber @xmath17 down in energy below the ones at zero 2d momentum@xcite .
this is confirmed by exact calculations of the two - orbital @xmath12-@xmath13 model with one electron more than half filling over a @xmath116 lattice of iron atoms under periodic boundary conditions . the previous schwinger - boson - slave - fermion description ( [ constraint ] ) for spin @xmath21 electrons is exploited to impose strong on - site - orbital coulomb repulsion .
details are given in ref .
@xcite .
figure [ exact]a shows the exact spectrum at the qcp , where @xmath94 .
the @xmath12-@xmath13 model parameters coincide with those set by fig .
[ meanfield ] , but with @xmath117 , and with hund coupling tuned to the critical value @xmath118 .
red states have even parity under orbital swap , @xmath119 , while blue states have odd parity under it .
notice that the lowest - energy doubly - degenerate states at wave number @xmath17 , which are spin-@xmath0 , lie @xmath120 in energy above the doubly - degenerate spin-@xmath0 groundstates at zero 2d momentum .
the latter states ( purple ) move up in energy off the fermi level set by the groundstates at csdw momenta as hund coupling falls below the critical value , and they become nearly degenerate with the former states in the absence of hund s rule .
this dependence on hund coupling is demonstrated by the inset to fig .
[ exact]a and by supplemental fig .
s3 .
the exact low - energy spectrum at sub - critical hund coupling is therefore consistent with the emergent hole bands obtained by the meanfield approximation , fig .
[ meanfield]b , but with the hole bands centered at wavenumber @xmath17 pulled down to much lower energy .
last , fig .
[ exact]a shows that the even parity ( @xmath14 ) and odd parity ( @xmath15 ) spin-@xmath0 groundstates at wavenumber @xmath121 are nearly degenerate , which suggests isotropic emergent hole bands at zero 2d momentum near the qcp . _
cooper pairs .
_ figure [ exact]b shows the spectrum of the same two - orbital @xmath12-@xmath13 model ( [ tj ] ) , but with two electrons more than half filling .
a _ repulsive _ interaction has been added to the heisenberg exchange terms in order to reduce finite - size effects : @xmath122 , equal to @xmath0 the spin - exchange operator . here
, @xmath123 counts the net occupation of _ holes _ per site - orbital . also , the on - site repulsion between mobile electrons in the @xmath70 and @xmath69 orbitals , respectively , is set to a large value @xmath124
. the schwinger - boson - slave - fermion description of the correlation electron ( [ constraint ] ) is again employed , with @xmath21 .
details are given in ref .
@xcite .
last , the ferromagnetic hund s rule exchange coupling constant is tuned to the critical value @xmath125 , at which @xmath94 .
this is depicted by the dashed horizontal line in fig .
[ exact]b , which shows the degeneracy between the csdw spin resonance at wavenumber @xmath6 with the hidden - order spin resonance at wavenumber @xmath17 .
the former is even ( black ) under swap of the orbitals , @xmath126 , while latter is odd ( red ) under it .
notice that the groundstate and the second excited state both lie under a continuum of states at zero net momentum .
they respectively have even and odd parity under a reflection about the @xmath127-@xmath54 diagonal .
we therefore assign @xmath128 symmetry to the groundstate bound pair and @xmath129 symmetry to the excited - state bound pair .
the dependence of the energy - splitting between these two states on hund coupling is shown by the inset to fig .
[ exact]b .
it provides evidence for a true qcp in the thermodynamic limit at @xmath130 , where the @xmath131-wave and @xmath5-wave bound states become degenerate .
figure [ s+-d+- ] depicts the order parameters for superconductivity of the two bound pair states shown in fig .
[ exact]b : @xmath132 times @xmath133 , with @xmath134 . here
, @xmath135 denotes the critical antiferromagnetic state of the corresponding heisenberg model@xcite at @xmath136 .
( see supplemental fig .
s4 . )
the groundstate has @xmath128 symmetry , as expected , but it also alternates in sign between cooper pairs at electron fermi surface pockets versus cooper pairs at the emergent hole bands .
( see fig .
[ meanfield]b . )
figure [ s+-d+- ] also shows that the ( second ) excited state has @xmath129 symmetry , as expected , and that it alternates in sign in a similar way .
the present exact results therefore provide evidence for remnant pairing on the emergent hole bands that lie below the fermi level at zero 2d momentum . , symmetrized with respect to both reflections about the principal axes . ]
_ discussion and conclusions .
_ the electronic structure in single - layer fese / sto is qualitatively described by the combination of figs .
[ meanfield]b and [ exact]a .
for example , a fit of inelastic neutron scattering data in iron - pnictide superconductors to the true linear spinwave spectrum fig .
[ meanfield]a , but at the qcp , yields @xmath137 mev , @xmath43 , and @xmath138 mev @xmath139 for the heisenberg exchange coupling constants@xcite .
hopping parameters set in figs .
[ meanfield]b and [ exact]a imply that the bottom of the electron bands lies @xmath140 mev below the fermi level . also , the csdw spin gap displayed by fig .
[ meanfield]a at sub - critical hund coupling is approximately @xmath10 mev , which therefore implies that the emergent hole bands at zero 2d momentum lie @xmath141 mev below the fermi .
both energy levels are roughly consistent with arpes in single - layer fese / sto@xcite . notably absent from our local moment model ( [ tj ] ) is the @xmath142 electron orbital of the iron atom .
dft calculations predict inner and outer electron fermi surface pockets at the corner of the two - iron brillouin zone that have @xmath143 and @xmath144 orbital character , respectively@xcite .
arpes and stm on heavily electron - doped surfaces of fese find an isotropic gap on the electron pockets , which rules out @xmath5-wave pairing symmetry@xcite . in such case ,
the limit of strong on - site coulomb repulsion assumed here would require remnant @xmath131-wave pairing of opposite sign on the buried @xmath144 band centered at zero 2d momentum .
the spectral weight of this band is negligibly small compared to that of the buried @xmath143 hole bands according to high - resolution arpes on alkali - metal dosed fese@xcite , however .
this contradiction argues that the iron @xmath142 orbital does not play an important role in high - temperature superconductivity shown at surface layers of heavily electron - doped fese .
last , fig .
[ s+-d+- ] predicts @xmath131-wave cooper pairs on the electron fermi surface pockets at csdw momenta .
this is again consistent with arpes and with stm on single - layer fese / sto , which find a gap at the fermi surface pockets , and no evidence for nodes@xcite .
figure [ s+-d+- ] also predicts remnant cooper pairs of opposite sign on the emergent hole bands that lie below the fermi level at zero 2d momentum .
the remnant pairs are likely a result of the intrinsic broadening in frequency ( [ intrinsic_broad ] ) experienced by the emergent holes .
( cf .
ref .
@xcite . )
recent quasi - particle interference patterns obtained from surface layers of intercalated fese observed a feature at csdw wavenumbers that could be accounted for by the superposition of an electron near csdw momenta with an andreev reflected hole near zero 2d momentum@xcite .
this work was supported in part by the us air force office of scientific research under grant no .
fa9550 - 13 - 1 - 0118 and by the national science foundation under prem grant no .
dmr-1523588 .
y. kamihara , t. watanabe , m. hirano , and h. hosono , j. am .
chem .
soc .
130 , 3296 ( 2008 ) .
q .- y .
wang , z. li , w .- h .
zhang , z .- c .
zhang , j .- s .
zhang , w. li , h. ding , y .- b .
ou , p. deng , k. chang , j. wen , c .-
l .
song , k. he , j .- f .
jia , s .- h .
ji , y. wang , l. wang , x. chen , x. ma , q .- k .
xue , chin .
. lett .
* 29 * , 037402 ( 2012 ) .
w .- h .
zhang , y. sun , j .- s .
zhang , f .- s .
li , m .- h .
guo , y .- f .
zhao , h .-
m .
zhang , j .-
p .
peng , y. xing , h .- c .
wang , t. fujita , a. hirata , z. li , h. ding , c .- j .
tang , m. wang , q .- y .
wang , k. he , s .- h .
ji , x. chen , j .- f .
wang , z .- c .
xia , l. li , y .- y .
wang , j. wang , l .- l . wang , m .- w .
chen , q .- k .
xue , and x .- c .
ma , chin .
phys .
lett . * 31 * , 017401 ( 2014 ) .
l.z .
deng , b. lv , z. wu , y.y .
xue , w.h .
zhang , f.s .
li , l.l .
wang , x.c .
ma , q.k .
xue , and c.w .
chu , phys .
rev .
b * 90 * , 214513 ( 2014 ) .
j .- f .
ge , z .-
l .
liu , c. liu , c .-
l .
gao , d. qian , q .- k .
xue , y. liu , j .- f .
jia , nature mat .
* 14 * , 285 ( 2015 ) .
d. liu , w. zhang , d. mou , j. he , y .- b .
ou , q .- y .
wang , z. li , l. wang , l. zhao , s. he , y. peng , x. liu , c. chaoyu , l. yu , g. liu , x. dong , j. zhang , c. chen , z. xu , j. hu , x. chen , z. ma , q. xue and x.j .
xhou , nature comm . * 3 * , 931 ( 2012 ) .
s. he , j. he , w .- h .
zhang , l. zhao , d. liu , x. liu , d. mou , y .- b .
ou , q .- y .
wang , z. li , l. wang , y. peng , y. liu , c. chen , l. yu , g. liu , x. dong , j. xhang , c. chen , z. xu , x. chen , x. ma , q. xue , and x.j .
zhou , nature materials * 12 * , 605 ( 2013 ) .
r. peng , x. p. shen , x. xie , h. c. xu , s.y .
tan , m. xia , t. zhang , h. y. cao , x. g. gong , j. p. hu , b. p. xie , d. l. feng , phys .
rev .
lett . * 112 * , 107001 ( 2014 ) .
q. fan , w. h. zhang , x. liu , y. j. yan , m. q. ren , r. peng , h. c. xu , b. p. xie , j. p. hu , t. zhang , and d. l. feng , nature physics * 11 * , 946 ( 2015 ) .
l. zhao , a. liang , d. yuan , y. hu , d. liu , j. huang , s. he , b. shen , y. xu , x. liu , l. yu , g. liu , h. zhou , y. huang , x. dong , f. zhou , z. zhao , c. chen , z. xu , x.j .
zhou , nat .
comm .
* 7 * , 10608 ( 2016 ) .
x.h .
niu , r. peng , h.c .
xu , y.j .
yan , j. jiang , d.f .
xu , t.l .
yu , q. song , z.c .
huang , y.x .
wang , b.p .
xie , x.f .
lu , n.z .
wang , x.h .
chen , z. sun , and d.l .
feng , phys .
rev .
b * 92 * , 060504(r ) ( 2015 )
. y. j. yan , w. h. zhang , m. q. ren , x. liu , x. f. lu , n. z. wang , x. h. niu , q. fan , j. miao , r. tao , b. p. xie , x. h. chen , t. zhang , d. l. feng , arxiv:1507.02577 . y. miyata , k. nakayama , k. suawara , t. sato , and t. takahashi , nat .
mater .
* 14 * , 775 ( 2015 ) .
c.h.p .
wen , h.c .
xu , c. chen , z.c .
huang , x. lou , y.j .
pu , q. song , b.p .
xie , m. abdel - hafiez , d.a .
chareev , a.n .
vasiliev , r. peng , and d.l .
feng , nat .
comm .
* 7 * , 10840 , ( 2016 )
. z.r .
ye , c.f .
zhang , h.l .
ning , w. li , l. chen , t. jia , m. hashimoto , d.h .
lu , z .- x .
shen , and y. zhang , arxiv:1512.02526 .
c .- l .
song , h .-
m .
zhang , y. zhong , x .- p .
hu , s .- h .
ji , l. wang , k. he , x .- c .
ma , and q .- k .
xue , phys .
rev .
lett . * 116 * , 157001 ( 2016 ) .
b. lei , j.h .
cui , z.j .
xiang , c. shang , n.z .
wang , g.j . ye , x.g .
luo , t. wu , z. sun , and x.h .
chen , phys .
rev .
lett . * 116 * , 077002 ( 2016 ) .
k. hanzawa , h. sato , h. hiramatsu , t. kamiya , and h. hosono , proc .
nat .
acad .
sci . * 113 * , 3986 ( 2016 ) .
t. bazhirov and m.l .
cohen , j. phys . : condens .
matter * 25 * , 105506 ( 2013 ) .
j. he , x. liu , w. zhang , l. zhao , d. liu , s. he , d. mou , f. li , c. tang , z. li , l. wang , y. peng , y. liu , c. chen , l. yu , g. liu , x. dong , j. zhang , c. chen , z. xu , x. chen , x. ma , q. xue , x. j. zhou , proc .
nat .
acad .
sci . * 111 * , 18501 ( 2014 ) .
q. si and e. abrahams , phys .
rev .
lett . * 101 * , 076401 ( 2008 ) .
j.p . rodriguez and e.h .
rezayi , phys .
rev .
lett . * 103 * , 097204 ( 2009 ) .
i.i .
mazin , d.j .
singh , m.d .
johannes , and m.h .
du , phys .
rev .
lett .
* 101 * , 057003 ( 2008 ) .
k. kuroki , s. onari , r. arita , h. usui , y. tanaka , h. kontani , and h. aoki , phys .
rev .
lett . * 101 * , 087004 ( 2008 ) .
j.p .
rodriguez , m.a.n .
araujo , p.d .
sacramento , phys .
rev .
b * 84 * , 224504 ( 2011 ) .
j.p . rodriguez , m.a.n .
araujo , p.d .
sacramento , eur .
phys .
j. b * 87 * , 163 ( 2014 ) .
d.p .
arovas and a. auerbach , phys .
rev .
b * 38 * , 316 ( 1988 ) .
c.l .
kane , p.a .
lee and n. read , phys .
rev .
b * 39 * , 6880 ( 1989 )
. a. auerbach and b. e. larson , phys .
rev .
b * 43 * , 7800 ( 1991 ) .
j.p .
rodriguez , phys .
rev .
b * 82 * , 014505 ( 2010 ) . see supplemental material at .... a. auerbach and d.p .
arovas , phys .
rev .
lett . * 61 * , 617 ( 1988 ) . at wavenumbers @xmath111 and @xmath17 ,
the spectral weight in ( [ emergent_holes ] ) diverges as the csdw spin gap vanishes at the qcp .
it yields the bound @xmath145 that guarantees the validity of ( [ emergent_holes ] ) . here
, @xmath146 coincides with the velocity of csdw spinwaves at the qcp@xcite .
the bound @xmath147 , likewise , guarantees the validity of ( [ emergent_holes ] ) at @xmath148 near csdw wavenumbers @xmath149 . at @xmath50 , the convolution ( [ g ] ) implies intrinsic broadening of the electron bands below the fermi level : @xmath150 at @xmath151 .
the goldstone mode associated with hidden magnetic order ( inset to fig . [ meanfield]a ) likely no longer exists once @xmath152 , however . in such case , the coherent and incoherent @xmath15 bands shown in fig .
[ meanfield]b may experience level repulsion at csdw momentum @xmath6 , likewise the @xmath14 bands at csdw momentum @xmath7 . in the independent electron approximation , adding small but positive intra - orbital nearest - neighbor hopping , @xmath153 , results in an electron band centered at wavenumber @xmath6 for the @xmath15 orbital with mass anisotropy @xmath154 .
the mass anisotropy of the corresponding emergent hole band at zero 2d momentum is the reverse within the schwinger - boson - slave - fermion meanfield approximation , at ideal hopping , @xmath50 .
adding the mean fields @xmath155 and @xmath156 for intra - orbital nearest - neighbor hopping of electrons ( before any @xmath157 degree spin rotations about the @xmath54 axis ) results in elliptical electron fermi surface pockets at wavenumbers @xmath6 and @xmath7 , respectively , for the @xmath15 and the @xmath14 orbitals , with the major axes along the principal axes . both of the extra mean fields vanish in the optimized theory at large @xmath75 , however .
( cf .
ref .
@xcite . )
o.k .
andersen and l. boeri , annalen der physik * 523 * , 8 ( 2011 ) .
x. chen , s. maiti , a. linscheid and p.j .
hirschfeld , phys .
rev .
b * 92 * , 224514 ( 2015 ) . *
supplemental material : isotropic cooper pairs with emergent sign changes in single - layer high-@xmath158 superconductor * _ department of physics and astronomy , _ _
california state university at los angeles , los angeles , ca 90032 _ |
a series of unexpected astrophysical observations have failed to find explanations in terms of standard astrophysical or particle physics models @xcite .
each of these anomalies can be explained , however , if there exists a dark weakly interacting massive particle , wimp , belonging to a secluded gauge sector @xcite . a dark vector boson ,
@xmath0 , an abelian gauge field , may couple the secluded sector to the standard model through its kinetic mixing with the standard model electroweak hypercharge gauge field , @xmath4 .
the kinetic mixing parameter , @xmath5 , is expected to be of the order @xmath6@xmath7 which allows for observable effects in @xmath8(gev)-energy @xmath9@xmath10 colliders @xcite .
the u boson might be produced in such collider experiments via several processes : @xmath11 decays , where @xmath12 and @xmath13 are vector and pseudoscalar mesons , @xmath2 with @xmath14 , where @xmath15 or @xmath16 , and @xmath17 ( dark higgsstrahlung ) , where @xmath18 is a higgs - like particle responsible for breaking the hidden symmetry .
the kloe experiment operated from 2000 to 2006 at da@xmath1ne , the frascati @xmath19 factory .
da@xmath1ne is an @xmath9@xmath10 collider running mainly at a center - of - mass energy of @xmath201.0195 gev , the mass of the @xmath19 meson .
equal energy electron and positron beams collide at an angle of @xmath2025 mrad , producing @xmath19 mesons nearly at rest .
the detector consists of a large cylindrical drift chamber ( dc ) @xcite , providing a momentum resolution of @xmath21 0.4 , surrounded by a lead - scintillating fiber electromagnetic calorimeter ( emc ) @xcite providing an energy resolution of @xmath22 and a time resolution of @xmath23 = 57 @xmath24 100 ps . a superconducting coil around the emc provides a 0.52 t field .
the kloe-2 collaboration has completed three searches for a dark photon .
the first two searched for u boson in vector meson decays @xmath25 , where @xmath26 , @xmath27 with the pseudoscalar meson decaying via @xmath28 @xcite and @xmath29 @xcite .
kloe-2 provided another limit for u boson production using the process @xmath30
, @xmath31 @xcite .
a fourth dark force analysis has been performed by kloe-2 by searching for the u boson in the dark higgsstrahlung process , @xmath17 . a preliminary limit on the product of the dark coupling strength and the kinetic mixing strength , @xmath32 , will be published soon .
the first three analyses produced excellent limits in the parameter space @xmath35 versus @xmath36 , but some values of @xmath5 and @xmath36 that can explain the @xmath37 anomaly have not yet been excluded . in particular
we would like to probe the range @xmath38 me-.07em v-.09em /-.08em c@xmath39 to either find evidence for an explanation of the muon anomaly or completely exclude the dark photon as a possible explanation . at an @xmath40 collider like da@xmath1ne , it s possible that the electron and positron can annihilate , or scatter , producing a u boson and a photon , with the decay of the u boson into a pair of leptons . unlike the previous kloe-2 limits , the sensitivity from the @xmath41 channel is expected to increase as @xmath36 aproaches @xmath42 due to the dramatic increase in the u boson production cross section , @xmath43 where we have electrons as our final - state leptons ( @xmath44 ) and @xmath45 is the total width .
a new kloe-2 analysis is underway which proposes to search for u boson production in the process @xmath30
, @xmath27 .
the 3 final - state particles of this process are the same as radiative bhabha scattering .
the distinct feature we are searching for is a breit - wigner resonant production peak ( at the u boson mass ) in the invariant - mass distribution of the @xmath40 pair . to search for
a u boson produced at a fixed - energy @xmath9@xmath10 collider we use initial - state radiation ( isr ) to reduce the center of mass energy and thereby scan the range of possible u boson masses down to @xmath42 .
the process consists of finite - width effects for @xmath46-channel annihilation subprocesses , non - resonant @xmath47-channel u boson exchange , and @xmath46-@xmath47 interference contributions .
the finite - width effects are order @xmath48 on the integrated cross section so are much smaller than any potential resonance we would observe , but they are critical from a phenomonological perspective and are properly taken into account in the monte carlo simulation @xcite .
the non - resonant @xmath47-channel effects would not produce the breit - wigner peak in the invariant mass distribution but could , in principle , show up in analyses of angular distributions or asymmetries .
the kloe-2 analysis will focus exclusively on resonant @xmath46-channel u boson production .
using about 1.5 fb@xmath49 of kloe data collected during 20042005 we will search for u boson production in a sample of radiative bhabha scattering events .
the strategy is to select events with the final - state electron , positron , and photon , all emitted at large angle ( @xmath50 ) with respect to the beam axis , such that they are explicitly detected in the barrel of the calorimeter .
the @xmath51 variable , computed using energy and momentum conservation , with the assumption of equal - mass oppositely - charged particles , will be used to separate electrons from the more massive muons and pions .
we will use monte carlo ( mc ) simulations to estimate the level of background contamination due to the following processes : @xmath52 , @xmath53 , @xmath54 ( where one photon converts into an @xmath40 pair ) , and @xmath55 , as well as other @xmath19 decays . due to the kloe detector s excellent efficiency at detecting electrons and distinguishing them from heavier charged particles
, we estimate that the sum of all background processes is typically less than 1% in the @xmath56 distribution .
none of the background shapes are peaked , eliminating the possiblility of a background mimicking the resonant u boson signal .
several monte carlo event generators for radiative bhabha scattering fail to accurately reproduce the physics at the dielectron mass threshold due to numerical instabilities in integrations of the form @xmath57 . due to the three order - of - magnitude difference between the electron mass and the center - of - mass energy of the collision , numerical instabilities arise as @xmath58 approaches threshold where the square root gives 0 , but as @xmath58 becomes larger than @xmath59 the factor @xmath60 becomes dominant .
these problems are apparent when the simulated cross section fails to show the significant rise at threshold . together with the authors of babayaga we modified the babayaga - nlo @xcite event generator and implemented it into our full kloe simulation such that the weighted events are distributed throughout the phase space with the square of the matrix element providing the correct weight . the good agreement between our mc simulation using the new event generator and our selected data sample
is shown in fig .
[ fig : m_ee_distribution ] .
no signal peak has been observed so far .
a preliminary excercise was performed on measured data using the cls technique @xcite to determine a preliminary limit on the number of signal u boson events , @xmath61 , at 90% confidence level .
chebyshev polynomials were fit to the measured data ( @xmath6215@xmath63 ) , excluding the signal region of interest ( @xmath623@xmath63 ) , and were used as the background .
a breit - wigner peak smeared with the invariant mass resolution was used as the signal .
we then translated this limit on @xmath61 to a 90% confidence level limit on the kinetic mixing parameter as a function of @xmath56 as @xcite @xmath64 where the radiator function @xmath65 was extracted from @xmath66 using the phokhara mc simulation @xcite to determine the radiative differential cross section , @xmath67 is the integral of the cross section ( [ eq : ubos_cross_sec ] ) , and @xmath68 fb@xmath49 is the integrated luminosity . the selection efficiency , @xmath69 , was obtained from a babayaga mc simulation where the radiative bhabha scattering was only allowed to proceed via the annihilation channel , since that is the channel in which the u boson breit - wigner resonance would occur ; the @xmath47-channel ultimately becoming a background .
our preliminary limit is shown in fig . [
fig : limits ] along with the limit from @xmath70 at 5@xmath63 , e141 @xcite , e774 @xcite , kloe(@xmath71 , @xmath3 ) @xcite , apex @xcite , wasa @xcite , hades @xcite , a1 @xcite , kloe(@xmath2 , @xmath72 ) @xcite , and a preliminary result from babar @xcite . .
we outlined our strategy for a new dark gauge u boson search in the process @xmath2 with @xmath3 using @xmath201.5 fb@xmath49 of kloe data collected in 20042005 .
after a preliminary excercise , we found no evidence for the existence of a u boson and set a preliminary upper limit at @xmath73@xmath74 on the level of kinetic mixing with the standard model as a function of the u boson mass in the range 10520 mev / c@xmath75 .
a final result is forthcoming and should extend the limit closer to the dielectron mass threshold .
the upgraded kloe-2 experiment @xcite , currently running , uses a new cylindrical gem inner tracker @xcite providing higher resolution interaction vertexing , and plans to collect upwards of 10 fb@xmath49 of data . the increased statistical power and tracking / vertexing sensitivity will allow kloe-2 to significantly extend our present limits .
we warmly thank our former kloe colleagues for the access to the data collected during the kloe data taking campaign .
we thank the da@xmath1ne team for their efforts in maintaining low background running conditions and their collaboration during all data taking .
we want to thank our technical staff : g.f .
fortugno and f. sborzacchi for their dedication in ensuring efficient operation of the kloe computing facilities ; m. anelli for his continuous attention to the gas system and detector safety ; a. balla , m. gatta , g. corradi and g. papalino for electronics maintenance ; m. santoni , g. paoluzzi and r. rosellini for general detector support ; c. piscitelli for his help during major maintenance periods .
this work was supported in part by the eu integrated infrastructure initiative hadron physics project under contract number rii3-ct- 2004 - 506078 ; by the european commission under the 7@xmath76 framework programme through the ` research infrastructures ' action of the ` capacities ' programme , call : fp7-infrastructures-2008 - 1 , grant agreement no .
227431 ; by the polish national science centre through the grants no .
dec-2011/03/n / st2/02641 , 2011/01/d / st2/00748 , 2011/03/n / st2/02652 , 2013/08/m / st2/00323 , and by the foundation for polish science through the mpd programme and the project homing plus bis/2011 - 4/3 . | dark matter and dark energy are two of the most fundamental open questions in physics today .
the existence of a light dark - force mediator has been hypothesized as a possible explanation for several unexplained physical phenomena .
a new search for this mediator , the dark photon @xmath0 , is underway using data collected with the kloe detector at da@xmath1ne .
we describe the strategy we will use in our search for a resonant peak in the electron - positron invariant mass spectrum from the process @xmath2 with @xmath3 .
so far we found no evidence for the process and set a preliminary upper limit on the level of mixing between the secluded dark sector and the standard model . |
magnetoelectric multiferroic materials which exhibit simultaneously ferroelectric and magnetic order are promising for new generation of random access memories ( ram ) , where the information can be written by electric field and read non - destructively by magnetic sensing . such memories avoid the weak points of the ferroelectric rams ( destructive reading causes fatigue ) as well as of magnetic rams ( high electric current is needed for overwriting , which rules out high integration of magnetic rams ) .
unfortunately , there are not many magnetoelectric multiferroics known up to now and only a few of them have both magnetic and ferroelectric critical temperatures above room temperature .
therefore there is nowadays an intensive search for magnetoelectric multiferroic materials with high magnetization and spontaneous polarization above room temperature.@xcite baettig and spaldin predicted from _ ab initio _ calculations that the chemically ordered double perovskite bi@xmath0fecro@xmath1 ( bfco ) will have - at zero temperature - a polarization of @xmath480@xmath5c/@xmath6 and a magnetization of @xmath4160 emu/@xmath7 ( 2 @xmath8 per formula unit).@xcite such properties far exceed the properties of any known multiferroic .
for the first time experimentally prepared epitaxial thin film of bfco which exhibited at room temperature ( rt ) a polarization of 2.8@xmath5c/@xmath6 and a saturated magnetization of 0.26@xmath8 per unit cell.@xcite recently kim et al.@xcite reported a remanent polarization of 60@xmath5c/@xmath6 at 77@xmath3 , for their bife@xmath9cr@xmath9o@xmath2 solid solution epitaxial filmsa nd alexe even measured 70 - 80@xmath5c/@xmath6 at rt on the films grown by nechache et al.@xcite the magnetic t@xmath10 and ferroelectric t@xmath11 phase transition temperatures in ordered bfco are not known up to now .
_ et al.__@xcite predicted from first principles a nel temperature t@xmath10 near 100@xmath3 , which was not confirmed in the experiments performed by nechache _
et al . _ , who observed magnetic order at rt .
very recently suchomel _ _
et al.__@xcite prepared bfco ceramics and observed a magnetic phase transition below 130@xmath3 , but it is worth noting that their ceramics exhibited chemical disorder of the fe@xmath12 and cr@xmath12 cations on the perovskite b site , which reduces t@xmath10 .
thus it is not excluded that t@xmath10 can be higher in ordered samples as in the case for the ordered bfco films reported by nechache at al.@xcite determination of the ferroelectric phase transition temperature is not possible from the low - frequency dielectric measurements due to the too high dc conductivity of the bfco film .
the extrinsic leakage conductivity does not play an appreciable role in the thz dielectric response of the film , therefore high - frequency dielectric studies are advantageous . for this purpose we performed infrared ( ir ) measurements including investigation in hardly achievable far infrared ( fir ) range below 200@xmath13 , which can give information about the complete phonon contributions to the static permittivity ( note that in the case of displacive ferroelectrics only polar phonons are responsible for the dielectric anomaly near t@xmath11 ) .
moreover , ir spectra usually change at the ferroelectric ( structural ) phase transition temperature due to the change of selection rules for ir active polar phonons .
therefore the ir spectra ( including fir ) of the bfco film can help to estimate its t@xmath11 as well as the symmetry of the high - temperature phase .
ir studies of ferroelectric thin films are rather rare in the literature and up to now almost only fir transmission spectra of the films deposited on fir - transparent substrates like si , sapphire or mgo were investigated .
fir transmission can give results only in a limited frequency range determined by the transparency window of the substrates , which is mostly very narrow , particularly at high temperatures ( e.g. sapphire is partially transparent only below 150@xmath13 at 900@xmath3).@xcite ir reflectance can yield results in a much broader spectral range , but its sensitivity is limited a ) by the thickness of the film , b ) by the strengths of polar phonons in the ir spectra and c ) by the ir properties of the substrate .
our experience shows that the substrates with buffer electrodes are not suitable due to the negative permittivity of the buffer layers , which reduces the sensitivity of the method .
therefore , dielectric substrates , which do not show any strongly temperature - dependent ir reflectivity spectra , are the most suitable for reflectance studies of thin films .
nevertheless , ir reflectance spectra of the thin films deposited on the substrate are strongly influenced by the substrate , since the thin films are partially transparent for the ir wavelength .
therefore both ir spectra of the bare substrate and of the film on the substrate should be measured at the same temperatures and the film properties
are evaluated from the spectra fits to such a multilayer system .
this method was used only twice in the literature for room or low - temperature ir studies of srtio@xmath2 films.@xcite here , we will use this method for the first time above room temperature and up to 900@xmath3 . in this paper
we shall show that the static permittivity of the bfco thin film , determined from the polar phonon contributions , increases monotonically on heating to 900@xmath3 , due to the phonon softening .
some phonon anomalies , probably connected with a magnetic phase transition , were observed near 600@xmath3 , but no dramatic changes , such as those usually related to a ferroelectric phase transition , were observed .
therefore , it seems that the phase transition to the paraelectric phase in bfco thin film occurs above the highest investigated temperature of 900@xmath3 .
we shall also report on our study of the magnetic properties of the bfco film up to 1000@xmath3 and we shall show that the magnetic phase transition occurs between 600 and 800@xmath3 .
bfco films were grown directly on ( 100)-oriented srtio@xmath2 substrates doped with 0.5 wt% of nb ( abbreviated sto : nb ) as well as on a ( 100)-oriented laalo@xmath2 substrate , better suited for ir measurements . an epitaxial 210 nm thick film deposited on the conducting sto : nb substrate was used for mssbauer spectroscopy and vibrating sample magnetometry studies .
xrd and tem data have shown that the thin film is epitaxial .
weak superlattice spots showed evidence of partial chemical order of the fe and cr cations.@xcite the srtio@xmath2 substrate exhibits a strongly temperature - dependent fir spectra due to the presence of an optic soft mode , which gives large inaccuracies in evaluation of the fir properties of the thin film .
we therefore studied fir spectra of the bfco film deposited on non - conducting laalo@xmath2 substrates with the size 5x10x0.5 mm .
since polar phonons are rather weak in bfco , we investigated a 600 nm thick film .
the relaxed film on laalo@xmath2 substrate ( orientation ( 001 ) ) was epitaxial with orientation ( 100 ) , but it revealed only very weak superlattice spots , so its chemical order in the b site was only partial .
xrd analysis of the film on laalo@xmath2 revealed a slight bi - deficiency in the bfco phase as well as about 5% of cr - doped bi@xmath0o@xmath2 secondary phase , which were not observed for the films on srtio@xmath2 .
the unpolarized fir and ir reflectance spectra were taken using a bruker ifs 113v ftir spectrometer at temperatures between 10 and 900@xmath3 with the resolution of 2@xmath13 .
an optistat cf cryostat from oxford instruments equipped with polyethylene windows was used for cooling the sample down to 10@xmath3 , while a commercial high temperature cell specac p / n 5850 was used for heating it up to 900@xmath3 .
a helium - cooled si bolometer operating at 1.6@xmath3 was used as a detector at low temperature measurements , while pyroelectric dtgs detectors were used for the ir measurements above rt .
magnetic properties of the bfco films on substrates 3x3 mm in size were investigated using a ppms 14 vibrating - sample magnetometer ( quantum design ) between 3 and 1000@xmath3 .
the mssbauer spectrum measurement was carried out in the conversion electron mssbauer spectroscopy ( cems ) mode with @xmath14co diffused into an rh matrix as a source moving with a constant acceleration .
the spectrum was accumulated for 7 days .
classical mssbauer spectroscopy in transmission mode could not be used due to the small volume of the investigated thin films .
the wissel spectrometer was calibrated by means of a standard @xmath15-fe foil , and the isomer shift was expressed with respect to this standard at 293@xmath3 .
the fitting of the spectra was performed using the normos program .
the cems method requires a conducting sample , therefore we investigated the thin film deposited on a conducting sto : nb substrate , while the films deposited on non - conducting laalo@xmath2 were more suitable for the ir studies .
fig . [ fig1 ] shows ir reflectance spectra of both a pure laalo@xmath2 substrate and a bfco thin film ( deposited on laalo@xmath2 ) at selected temperatures between 10 and 900@xmath3 . only small temperature dependence of the reflectivity spectra of the laalo@xmath2 substrate
can be seen , mostly due to an increase in phonon damping with temperature .
also the sharp peaks near 500 and 600@xmath13 gradually disappear due to a second order structural phase transition in laalo@xmath2 from trigonal to cubic phase at 800@xmath3.@xcite substrate and ( b ) reflectance spectra of bi@xmath0fecro@xmath1 film ( 600 nm thick ) deposited on laalo@xmath2 .
the arrows show anomalous phonons .
note that 1@xmath13 corresponds to 30ghz.,width=302 ] ir reflectance spectra of the semitransparent bfco film deposited on the opaque laalo@xmath2 substrate exhibits more pronounced changes with temperature mainly near 100 , 250 and 550@xmath13 ( marked by arrows in fig . [ fig1 ] ) .
for the detailed analysis we evaluated the complex permittivity @xmath16 spectra of the film ( see fig.[fig2 ] ) using the following procedure : the reflectivity r(@xmath17 ) of the bare substrate at each temperature was first fitted using the formula @xmath18 where for the @xmath19(@xmath17 ) the factorized form of the complex permittivity @xcite was used @xmath20 @xmath21 and @xmath22 denote the frequencies of the j - th transverse and longitudinal polar phonon , respectively , and @xmath23@xmath24 and @xmath23@xmath25 denote their corresponding damping constants .
the high - frequency permittivity @xmath26 results from the electron absorption processes and from the phonon contribution above 600@xmath13
. then the spectrum of the two - slab system ( film + substrate ) was fitted using the full formula for the coherent reflectance of a two - layer system@xcite where the oscillator parameters of the substrate were fixed , in order to determine the oscillator parameters of the polar phonons in the film .
we note that for the film ir spectra fits we used the classical lorentz model of the damped harmonic oscillators instead of eq .
( [ eps4p ] ) @xmath27 where @xmath28 means the contribution of the j - th mode to the static permittivity .
the rest of the parameters in eq .
( [ eps3p ] ) have the same meaning as in eq .
( [ eps4p ] ) .
( [ eps4p ] ) is more suitable for the reflectivity fits of phonon spectra with a large to - lo splitting , when both kinds of phonon modes have different damping .
such a model was necessary to use for a good fit of the laalo@xmath2 substrate .
( [ eps3p ] ) is more appropriate for fitting of the reflectivity spectra of phonons with a small to - lo splitting and/or transmission spectra which do not show up anomalies at lo frequencies .
it has fewer parameters and gives acceptable physical results , while the former model can sometimes yield un - physical negative dielectric losses , when the parameters are not properly chosen .
complex dielectric spectra of bfco film obtained from the above described fit of the ir reflectance spectra displayed on fig .
[ fig1 ] are plotted in fig .
[ fig2 ] . the temperature dependence of to phonon frequencies is plotted in fig .
one can clearly see the shift of most of the phonon modes to lower frequencies on heating ( phonon softening ) .
it causes the gradual increase of the static permittivity with rising temperature ( see inset in fig .
[ fig2 ] ) . .
frequencies of the peaks in @xmath29(@xmath17 ) spectra roughly correspond to the to phonon frequencies .
note the continuous increase of the static permittivity @xmath30(0 ) with rising temperature ( see inset).,width=302 ] .,width=302 ] six modes ( respectively seven below 200@xmath3 ) were necessary for the fits .
let us compare the number of observed polar modes with the prediction of factor group analysis .
bfco crystallizes in the rhombohedral space group @xmath31 .
bi , fe and cr cations have the site symmetry @xmath32(1 ) , while oxygen ions have the @xmath33 site symmetry.@xcite factor group analysis of the lattice vibrations based on tables published by rousseau et al .
@xcite yields the following optic phonons @xmath34 it means that the 9@xmath35 and 9@xmath36 modes are both raman and ir active .
the analysis gives also additional 1@xmath35 + 1@xmath36 acoustic modes .
the modes with the @xmath35 symmetry are active in spectra with the electric vector * e * of the ir wave parallel to the @xmath37 axis , while the @xmath36 modes are active in * * e**@xmath38 @xmath39 spectra . the rest of symbols ( @xmath40 , @xmath41 etc . ) in brackets of the group analysis in eq .
[ eq : rhombo ] shows components of raman tensors , in which the phonons are raman active .
our epitaxial bfco film is ( 001 ) oriented and since we measure the in - plane response , we see mostly the @xmath36 symmetry modes in our fir spectra .
we resolved 7 modes in the low - temperature spectra although 9 @xmath36 modes are allowed .
this is quite reasonable , if we take into account that some of the modes have small intensity or they may overlap with other modes .
it is worth to note that the to phonon frequencies in bfco correspond very well to the @xmath36 symmetry to phonon frequencies in chemically and structurally related bifeo@xmath2.@xcite one mode near 50@xmath13 disappears from the fir spectra above 200@xmath3 ( see black solid squares in fig .
[ fig3 ] ) .
such change could be a hint of some structural phase transition , but the fir spectra near 50@xmath13 are rather noisy at high temperatures , so we can not exclude that the mode is present in the spectra also at higher temperatures , but we do not resolve it due to lower sensitivity of the high - temperature fir experiment .
the absence of any other phonon anomalies at higher frequencies also does not give some phase transition near 200@xmath3 .
interesting phonon anomalies are seen near 600@xmath3 ( see fig . [ fig3 ] ) .
some mode frequencies show relatively large temperature changes and splitting of the modes near 490 and 550@xmath13 that almost disappears above 600@xmath3 . however , it is important to stress that no mode from the doublet disappears above 600@xmath3 .
both modes remain in the spectra with similar frequencies near 520@xmath13 up to 900@xmath3 . suchomel _ at al._,@xcite observed decomposition of bfco ceramics on heating above 400@xmath42c
. this effect could be also responsible for the phonon anomalies seen near 600@xmath3 in our film , but we have to emphasize that we did not observe any decomposition in our sample ( placed in a vacuum chamber of the spectrometer ) , because the ir spectra and magnetic properties ( see below ) were reproducible before and after the thermal cycling .
phonon anomalies near t@xmath10=640k , similar to ours in fig .
[ fig3 ] , were observed in raman spectra of bifeo@xmath2 @xcite and they were explained by spin - phonon coupling .
we will discuss this possibility below together with the magnetic data .
the phonon frequency changes seen near 600@xmath3 can be a consequence of some phase transition , but probably not a ferroelectric one because we see a gradual increase of the static permittivity @xmath30(0 ) ( from phonon contributions ) on heating ( see inset in fig .
[ fig2 ] ) , while for a ferroelectric transition a maximum in @xmath43(@xmath44 ) should be seen near t@xmath11 .
it seems that the ferroelectric phase transition in bfco lies above 900@xmath3 , like for bifeo@xmath2 .
note that bifeo@xmath2 has a rhombohedral @xmath45 structure with a structural phase transition ( according to earlier literature ) to cubic paraelectric @xmath46 phase near 1120@xmath3 .
very recent structural studies@xcite revealed an intermediate orthorhombic phase with the space group @xmath47 or @xmath48 at temperatures between @xmath41100 and @xmath41200@xmath3 and probably only above @xmath41200@xmath3 bifeo@xmath2 transforms into the cubic and simultaneously conducting phase.@xcite if we assume that the structural phase sequence in bfco is the same as in bifeo@xmath2 , than the following factor group analysis of the optic phonons applies in the orthorhombic phase @xmath49 it means that instead of the 18 modes in the rhombohedral structure , 23 ir active modes ( 13 in * * e**@xmath38 @xmath39 ) should be seen in the ir spectra of the orthorhombic phase .
unfortunately , no new mode appears in our fir spectra , so we do not see any evidence for a phase transition into the orthorhombic phase at temperatures below 900@xmath3 .
on the other hand , in cubic paraelectric phase the following optic modes are expected : @xmath50 it seems that only 4 phonons of @xmath51 symmetry should be seen in the fir spectra and 4 modes ( @xmath52 , @xmath53 and @xmath54 symmetries ) in raman spectra .
we see 6 modes in fig .
it means that bfco probably remains in the rhombohedral phase in the whole investigated temperature range up to 900@xmath3 and the structural phase transition only occurs , similarly as in bifeo@xmath2 , at higher temperatures .
the absence of a phase transition from the ferroelectric to paraelectric phase below 900@xmath3 is also supported by observed gradual increase of the static permittivity on rising temperature - see inset in fig .
further investigations , like high - temperature structural or second harmonic generation , are needed for revealing the t@xmath11 and symmetry of the high - temperature phase(s ) . .
inset shows the room - temperature cems mssbauer spectrum of the same film together with its fit.,width=321 ] let us discuss the magnetic properties of bfco thin film , which we investigated by means of vibration magnetometry and cems mssbauer spectroscopy .
the bfco film deposited on the laalo@xmath2 substrate , originally investigated by ir spectroscopy , exhibits a strong diamagnetic signal from the substrate and the magnetic hysteresis loops were only revealed at low temperatures below 20@xmath3 ( not shown here ) .
it can be explained by a weak b - site order which suppresses the magnetic phase transition temperature . on the other hand , the well - ordered bfco thin film on the sto :
nb substrate exhibits nice magnetic hysteresis loops not only at rt but also at higher temperatures ( see fig .
[ fig4 ] ) . negative slope of magnetization at higher magnetic fields seen above 800@xmath3 can be explained by a diamagnetic contribution of the sto : nb substrate , but below 600@xmath3 the open hysteresis loop is clearly seen .
the value of saturated magnetization is typical for antiferromagnets with weak ferromagnetism induced by a canted spin structure and the value of rt spontaneous magnetization corresponds well to the previously published results.@xcite the low value ( @xmath4 0.3 @xmath8/f.u . ) of magnetization at saturation of the film regarding the expected theoretical value of 2 @xmath8/f.u .
@xcite could be explained by ( i ) the fe - cr ordering which may be only partial , ( ii ) the partial chemical disorder that generates an antiferromagnetic antisites contribution ( fe - fe , cr - cr ) , or / and ( iii ) the partial relaxation of the strain in the film leading to a more distorted structure .
[ fig5 ] shows the temperature dependence of the magnetization at various magnetic fields .
the magnetic experiments above and below 300@xmath3 were performed separately , which is probably the reason of the change of slope seen at 300@xmath3 in temperature dependence of magnetization measured at 1500oe .
the magnetization remains nonzero up to 1000@xmath3 in the field of 1500oe , but the hysteresis loop is very slim at temperatures above 800@xmath3 .
it is difficult to determine exactly the magnetic phase transition from fig .
[ fig4 ] and [ fig5 ] , but it seems that it could lie somewhere between 600 and 800@xmath3 .
it is worth noting that beattig _ _
et al.__@xcite predicted @xmath55 in bfco only near 100@xmath3 , which can correspond to recent result of suchomel et al .
@xcite , who claim , based on transmission mssbauer spectrum of bfco ceramics , that @xmath55 lies below 130@xmath3 .
low @xmath55 can be explained by chemical disorder in the fe and cr cations in the [ 111 ] direction,@xcite while our thin film is at least partially chemically ordered ( we observed satellites peaks both in the xrd and in the selected area electron diffraction patterns taken by tem ) .
the inset of fig .
[ fig4 ] shows the room - temperature cems mssbauer spectrum of the bfco film .
surprisingly , only a doublet is seen , which is typical for the _ paramagnetic _ state of fe@xmath12 ions in the sample , while a sextet is expected in a magnetically ordered state .
this is rather puzzling because clear magnetic hysteresis loops are seen in the same sample by vibration magnetometry ( fig .
[ fig4 ] ) .
conversion electron mssbauer spectroscopy of @xmath14fe is based on the detection of electrons with the energy of 7.3 kev which were knocked out of the k shell of the @xmath14fe atom after re - emission of the gamma quantum originally resonantly absorbed by the @xmath14fe nucleus .
the release is almost instantaneous ( within 10@xmath56 s ) and has a rather high probability .
most of these electrons are again absorbed in the material , but some of them , depending on the depth , where the electron emission occurs and on the electron work function , reach the surface of the sample and are finally detected .
the depth from which the information is collected by the cems method is usually @xmath4 200 nm depending on the absorption properties of the material , the electrons emitted from deeper regions of the sample do not reach the surface .
our thin film is only 210 nm thick , which means that we should see the cems signal from the whole volume of the film .
however , the film is not only magnetic but also ferroelectric , and electric field in the ferroelectric domains should substantially influence the work function of the electrons .
the ferroelectric domain structure of bfco is complex and assuming it is similar to that reported for bifeo@xmath2,@xcite the polarization ( and related internal electric field ) is oriented 41.8@xmath57 or even 131.8@xmath57 to the normal surface of ( 001 ) oriented thin film .
therefore , the emitted electrons are returned back to the film and most of them lose their energy , are absorbed and do not leave the film .
only electrons emitted from a very thin surface layer ( @xmath4 10 nm ) may reach the surface and are detected in the cems experiment .
the thin film surface layer is most likely non - magnetic ( probably due to chemical disorder of fe and cr cations at the surface ) , therefore only a doublet is observed in our cems mssbauer spectra shown in the inset of fig .
[ fig4 ] , although the volume of the film is magnetically ordered , as seen from the magnetic hysteresis loops measured by vibrating - sample magnetometry .
nevetheless , we have to stress that we repeated the cems mssbauer experiment also with another bfco film ( thickness 86 nm , sto : nb substrate ) , which exhibited strong satellites in the xrd ( i.e. a higher fe and cr chemical order than in the previous sample ) , as well as broad magnetic hysteresis loop and still we found only doublet in cems spectra typical for the paramagnetic order . finally , we note that the doublet can not originate from the substrate , because the mssbauer spectrum is sensitive only to the fe cations not present in the sto : nb substrate .
.comparison of the fit parameters of mssbauer spectra in fig .
[ fig1 ] and in ref .
[ cols= " < , < , < , < " , ] [ moesbauer ] parameters of the mssbauer spectra fit are summarized in table [ moesbauer ] . from the fe isomer shift @xmath58 the valency of the iron
can be clearly estimated .
the fe@xmath12 cations in the oxidic compound have its isomer shift in the range of 0.1 - 0.5 mm / s while the fe@xmath59 cations show @xmath58 in the range of 0.8 - 1.5 mm / s.@xcite our obtained value @xmath58=0.39mm / s confirms the absence of fe@xmath59 states and the presence of only fe@xmath12 states in the investigated film .
mssbauer spectra also allow determining the site symmetry for fe@xmath12 cations in the structure . according to refs.@xcite ,
the usual isomer shift values for fe@xmath12 in the case of the spectra measured at rt are as follows : 0.10 - 0.30 mm / s for fe@xmath12 in a tetrahedral site , 0.28 - 0.50 mm / s for fe@xmath12 in an octahedral site .
when we compare the above - mentioned ranges with our @xmath58=0.39 mm / s , we can confirm that fe@xmath12 in bfco is in the octahedral position .
when we compare our fitting parameters in table [ moesbauer ] with the parameters obtained from the mssbauer spectra ( measured in the transmission mode ) of disordered ceramics published by suchomel _ at al._,@xcite , we can state that both ceramics and surface layer of our thin film have the same or similar non - magnetic structure at rt , although the magnetic measurements by vibrating - sample magnetometry give evidence for a magnetic order in the thin film far above rt .
in the light of our high - temperature magnetic data we can suggest that the phonon anomalies seen near 600@xmath3 are due to a magnetic phase transition .
near this temperature a sudden drop of the permittivity is seen , which is typical for spin - phonon coupling.@xcite nevertheless , further magnetic , structural and dielectrics studies are necessary for the confirmation of this suggestion .
the complex dielectric response of a bfco film was determined by a novel method - ir reflectance of a bfco thin film deposited on a laalo@xmath2 substrate .
most of the polar phonons seen in the ir spectra reveal gradual softening on heating from 20 to 900@xmath3 , which causes a progressive increase of the static permittivity with increasing temperature .
therefore we speculate that the ferroelectric phase transition lies ( similarly to the related bifeo@xmath2 ) above 900@xmath3 , although some phonon anomalies , probably connected with the magnetic phase transition , were observed near 600@xmath3 .
magnetic properties of the bfco thin film were investigated between 6 and 1000@xmath3 and revealed that the bfco film is a good high - temperature multiferroic with a magnetic phase transition that we assumed to take place somewhere between 600 and 800@xmath3 .
conversion electron mssbauer spectrum did not reveal magnetic order in the bfco thin film in contrast to vibration magnetometry , because the electric field presented in ferroelectric domains extends the track of emitted electrons and prevents their detection from most the volume depth of the thin film .
therefore probably only electrons from the thin skin non - magnetic layer of the film are detected .
further structural , magnetic and dielectric high - temperature studies on a well b - site - ordered samples are in progress .
the work was supported by the grant agency of the czech republic ( project no .
202/06/0403 ) and grant agency of academy of science of the czech republic ( project no .
kjb100100704 ) and avoz10100520 .
part of this research was supported by grants from the natural sciences and engineering research council of canada ( nserc ) as well as from the fond qubcois de la recherche sur la nature et les technologies ( fqrnt ) .
the authors thank j. ebek for valuable discussions as well as j. petzelt and t.w .
johnston for critical reading of the manuscript .
m. fiebig , j. phys .
, * 38 * , r123 ( 2005 ) .
cheong and m. mostovoy , nature mater .
, * 6 * , 13 ( 2007 ) .
p. baettig and n.a .
spaldin , appl .
* 86 * , 012505 ( 2005 ) .
r. nechache , c. harnagea , a. pignolet , f. normandin , t. veres , l .- p .
carrignan , and d. mnard , appl .
. lett . * 89 * , 102902 ( 2006 ) .
r. nechache , c. harnagea , l .- p .
carrignan , d. mnard , and a. pignolet , philos . mag .
lett . * 87 * , 231 ( 2007 ) .
kim , h.n .
lee , m.d .
biegalski and h.m .
schristen , appl .
91 * , 042906 ( 2007 ) .
m. alexe and l. pintilie , private communication .
p. baettig , c. ederer and n.a .
spaldin , phys .
b * 72 * , 214105 ( 2005 ) . m.r .
suchomel , ch.i .
thomas , m. allix , m.j .
rosseinsky , a.m. fogg , and m.f .
thomas , appl .
. lett . * 90 * , 112909 ( 2007 ) .
r. nechache , l .- p .
carignan , l. gunawan , c. harnagea , g. a. botton , d. mnard , a. pignolet , j. mater . res . *
22 * , 2102 ( 2007 ) .
s. kamba , m. kempa , v. bovtun , j. petzelt , k. brinkman and n. setter , j. phys . : condens
. matter , * 17 * , 3965 ( 2005 ) .
almeida , a. pietka , p. caldelas , j.a .
mendes , j.l .
ribeiro , thin solid films , * 513 * 275 ( 2006 ) .
t. yamada , j. petzelt , a.k .
tagantsev , s. denisov , d. noujni , p.k .
petrov , a. mackova , k. fujito , t. kiguchi , k. shinozaki , n. mizutani , v.o .
sherman , p. muralt , and n. setter , phys .
lett . * 96 * , 157602 ( 2006 ) .
mueller , w. berlinger , f. waldner , phys .
* 21 * , 814 ( 1968 ) .
f. gervais , in infrared and millimeter waves , vol . 8 , ed . k. j. button ( new york : academic ) chapter 7 , p. 279 .
m. born and e. wolf , principles of optics , oxford , pergamon , 1960 .
o. s. heavens , rep .
* 22 * , 1 ( 1960 ) .
rousseau , r.p .
bauman , and s.p.s .
porto , j. raman spectroscopy * 10 * , 253 ( 1981 ) .
s. kamba et al .
b * 75 * , 024403 ( 2007 ) .
p. hermet , m. goffinet , j. kreisel , and ph .
ghosez , phys .
b * 75 * , 220102(r ) ( 2007 ) . r.p.s.m .
lobo , r.l .
moreira , d. lebeugle , d. colson , phys .
b * 76 * , 172105 ( 2007 ) .
r. haumont , j. kreisel , p. bouvier , and f. hippert , phys .
b * 73 * , 132101 ( 2006 )
scott , invited talk at 11th european meeting on ferroelectricity , bled , slovenia , september 2007 .
y .- h . chu , l.w .
martin , m.b .
holcomb , and r. ramesh , mat . today
* 10 * , 16 ( 2007 )
. f. menil , j. phys .
solids , * 46 * , 763 ( 1985 ) .
j. parmentier , s. vilminot , j .- l .
dormann , sol .
state sciences * 1 * , 257 ( 1999 ) .
smolenskii and i. chupis , sov .
* 25 * , 475 ( 1982 ) . | infrared reflectance spectra of an epitaxial bi@xmath0fecro@xmath1 thin film prepared by pulsed laser deposition on laalo@xmath2 substrate were recorded between 10 and 900@xmath3 .
no evidence for a phase transition to the paraelectric phase was observed , but some phonon anomalies were revealed near 600@xmath3 .
most of the polar modes exhibit only a gradual softening , which results in a continuous increase of the static permittivity on heating .
it indicates that the ferroelectric phase transition should occur somewhere above 900@xmath3 .
magnetic measurements performed up to 1000@xmath3 revealed a possible magnetic phase transition between 600 and 800@xmath3 , but the exact critical temperature can not be determined due to a strong diamagnetic signal from the substrate . nevertheless , our experimental data show that the b - site ordered bi@xmath0fecro@xmath1 is one of the rare high - temperature multiferroics . |
a dynamical system , which we deal with , is @xmath12 where @xmath1 is a solution ( _ wave _ ) , @xmath2 is a _ control_. the system describes propagation of incoming spherical waves initiated by `` infinitely far sources '' ( controls ) . the relevant control problem is : given @xmath13 to find @xmath14 obeying @xmath15 . with regard to hyperbolicity of the system , this problem can be specified as follows .
let @xmath16(@xmath17 ) be the subspace of delayed controls ( @xmath18 is the value of delay , @xmath19 )
. with the system one associates a family of _ reachable sets _ governing its evolution does not depend on time . by this ,
delay of control leads to the same delay of wave . ]
@xmath20 denote @xmath21 , @xmath22 .
the waves , which are governed by the wave equation ( [ eq3 ] ) , propagate with velocity @xmath23 . at the moment
@xmath24 , the incoming waves produced by delayed controls @xmath25 fill up the part @xmath26 of the space ( see ( [ eq2 ] ) ) .
therefore , the embedding @xmath27 holds .
the latter motivates the following setup of the _ local control problem _ ( lcp ) : given @xmath28 to find @xmath29 obeying @xmath15 . to study
its solvability is to clarify the character of the embedding ( [ u xi subseth xi ] ) .
in particular , the question is whether the delayed waves constitute an @xmath30-complete system in @xmath31 . for @xmath32
, the lax - phillips theory @xcite provides @xmath33 ( see also @xcite ) . in the mean time , as was first found out in @xcite , for @xmath34 the embedding ( [ u xi subseth xi ] ) is not dense : @xmath35 i.e. , a lack of local controllability does occur .
elements of @xmath7 are interpreted as _ locally _ unreachable / unobservable states of system ( [ eq1])([eq3 ] ) .
some of their properties are studied in @xcite@xcite . :
see @xcite , @xcite ] the subject of the paper is a smoothness of unobservable states .
in sec 3 , we provide the arguments connected with propagation of singularities ( geometrical optics ) , which motivated our starting conjecture : elements of @xmath7 should be smooth at least with respect to the radial variable @xmath36 . surprisingly , it turns out to be wrong : we construct a counterexample of @xmath37 such that @xmath38 .
it is this construction , which is our main result .
@xmath39 in sec 2 , we discuss an lcp for a bounded domain ( a ball ) .
the goal is to emphasize the principal difference : in a ball , for the straightforward analog of system ( [ eq1])([eq3 ] ) one has @xmath40 ,
i.e. the local controllability does hold . in the mean time , system ( [ eq1])([eq3 ] ) in the space ( and the lcp for it ) is a relevant limit case of the system in a ball .
so , a notable effect occurs : propagating from infinity , the waves lose completeness in the filled domains .
@xmath39local completeness of waves produced by boundary controls plays a crucial role in the _ boundary control method _ , which is an approach to inverse problems based on their relations to control theory @xcite . planning to apply the bc - method to the dynamical inverse scattering problem for the acoustical equation @xmath41 in @xmath42
, we have to bear in mind ( [ d^xi not= 0 ] ) as a fact , which complicates such an application .
it is the reason , which motivates our interest to unobservable states .
@xmath39the work is supported by the grants rfbr 11 - 01 - 00407a and spbgu + 11.38.63.2012 , 6.38.670.2013 .
denote @xmath44 ; fix a positive @xmath45 . here
we deal with a dynamical system @xmath43 of the form @xmath46
\times \gamma^t\ , , \end{aligned}\ ] ] where @xmath1 is a solution ( _ wave _ ) , @xmath47;l_2\left(\gamma^t\right)\right)$ ] is a _ control_.
the following is the standard control theory attributes of the system .
@xmath39the _ outer space _ is @xmath48 .
it contains the subspaces @xmath49\ } , \qquad 0\leqslant \xi \leqslant t\ ] ] ( @xmath50 ) , which consist of the delayed controls , @xmath18 being the delay , @xmath51 the action time .
@xmath39the _ inner space _ of states @xmath52 contains the subspaces @xmath53 ( @xmath54 ) , where @xmath55 .
@xmath39the ` input @xmath56 state ' correspondence is realized by a _ control operator _ @xmath57 .
this operator is bounded @xcite .
the waves governed by equation ( [ eq1 t ] ) propagate in @xmath58 with velocity @xmath23 .
delay of control implies the same delay of wave . as a result
, one has @xmath59 so that at the final moment the delayed waves fill up the near - boundary layer of the thickness @xmath51 .
introduce the _ reachable sets _ @xmath60 by ( [ supp u ] ) , the embedding @xmath61 holds .
the well - known fact is that this embedding is dense : @xmath62 ( the closure in @xmath63 ) .
it is derived from the fundamental holmgren - john uniqueness theorem ( see , e.g. , @xcite , chaper iv , theorem 1.5 ) by the scheme originated by d.russell in @xcite .
thus , the waves do constitute complete systems in the filled domains @xmath64 , i.e. , system @xmath43 is locally controllable from the boundary .
in addition , note that for @xmath65 one has @xmath66 , so that the reachable set is dense in the subspace @xmath67 but does not exhaust it @xcite .
therefore , one should speak about _ approximate _ local boundary controllability .
a dynamical system dual to @xmath43 is @xmath69 \times \gamma^t\ , , \end{aligned}\ ] ] where @xmath70 and @xmath71 is a solution . with this system one
associates an _ observation operator _
@xmath72 , @xmath73 where @xmath74 .
integration by parts leads to the relation @xmath75 which motivates the term ` dual ' .
elements of @xmath76 are said to be _
unobservable_. since @xmath77 there are no unobservable elements in the dual system .
now , fix a positive @xmath78 . in ( [ eq2 t * ] ) ,
let us take @xmath79 provided @xmath80 ( so that @xmath81 vanishes near @xmath82 ) , @xmath83 , and @xmath84 by standard propagation of singularity arguments ( see , e.g. , @xcite ) , the latter yields @xmath85 where @xmath86 are the characteristic cones ( fig 1a ) .
the cone @xmath87 supports the singularities of @xmath88 , which move ( in the reversed time ) towards the boundary @xmath89 and reach it at the moment @xmath90 .
the singularities reflected from the boundary are supported in @xmath91 . as a result , for the trace of the radial derivative @xmath92 at the boundary ( as a function on @xmath93 ) one has @xmath94 so , _ any _ ` spherical ' singularity ( [ sing supp y ( ball ) ] ) manifests itself at the boundary at the proper moment @xmath90 .
as example , consider @xmath81 with a radial jump .
denote @xmath95 , so that @xmath96 and @xmath97 .
assume that @xmath98 \,=\,\alpha(\omega ) , \quad \omega \in s^2 .
\end{aligned}\ ] ] with a nonzero @xmath99 .
then , @xmath100 and @xmath92 have a jump supported on the cones @xmath101 .
the jump at @xmath87 can be treated by standard geometrical optics considerations , which provide @xmath102\,\\ \label{jump d / dr v^y ( ball ) } & = \,- \frac{\xi_0}{\xi_0+t - t } \alpha(\omega ) , \qquad \omega \in s^2,\,\,\xi_0<t < t \end{aligned}\ ] ] ( see , e.g. , @xcite , chapter 3 ) . as a result , the trace of @xmath92 on the boundary turns out to be discontinuous at @xmath90 , and quite standard technique @xcite provides @xmath103\,=\,- \frac{\xi_0}{t } \alpha(\omega ) , \quad \omega \in s^2 .
\end{aligned}\ ] ] the latter is equivalent to @xmath104\big|_{x = t\omega}\,=\,- \frac{\xi_0}{t } \alpha(\omega ) , \quad \omega \in s^2 .
\end{aligned}\ ] ] recall that here @xmath105 is an @xmath106-valued function of @xmath107 $ ] , so that its jump at @xmath90 is an element of @xmath106 , its value being taken at the point @xmath108 .
this jump is observable in any time interval @xmath109 $ ] as @xmath110 .
here we deal with the system @xmath111 of the form @xmath112 where @xmath1 is a solution ( _ wave _ ) ; @xmath113 is a _ control _ , which we regard as an @xmath114-valued function of @xmath115 .
for a smooth enough @xmath113 vanishing near @xmath116 , problem ( [ eq1 again])([eq3 again ] ) has a unique classical solution , and the well - known representation @xmath117 holds , where @xmath118 is extension of @xmath113 from @xmath119 to @xmath120 _ by zero _ , @xmath121 , @xmath122 is the standard inner product in @xmath42 , and @xmath123 is the surface element on @xmath124 ( see , e.g. , @xcite ) .
@xmath39the ` input @xmath56 state ' correspondence is realized by a _ control operator _ @xmath132 . by ( [ u via f ] ) , for available controls one has @xmath133 moreover , @xmath134 is a unitary operator , @xcite , @xcite ) , so that @xmath135 holds ( see , e.g. , @xcite .
system @xmath111 can be regarded as a limit case of system @xmath43 . and @xmath136 , @xmath137 the solution to ( [ eq1t])([eq3 t ] ) .
then quite simple considerations provide @xmath138 ] however , they substantially differ in controllability properties . in system
@xmath111 , a relevant analog of the reachable sets @xmath139 of @xmath43 is @xmath140 since @xmath134 is a unitary operator , @xmath141 is a closed subspace . for @xmath65 in a ball ] the equality @xmath142 is easily seen from ( [ eq2 again ] ) , ( [ u via f ] ) and corresponds to the steady - state property .
also , for a delayed @xmath25 , ( [ repres w ] ) evidently implies @xmath143 , i.e. , @xmath144 thus , at the final moment @xmath24 , the delayed waves incoming from infinity fill up the domain @xmath145 , and the embedding @xmath146 holds .
the question is whether they are complete in @xmath31 , i.e. , the local controllability occurs .
the answer is negative : as was found out in @xcite , @xcite , in contrast to ( [ u xi = h xi ] ) , one has @xmath147 in the mean time , for @xmath32 the equality @xmath148 does hold .
elements of @xmath7 are said to be the states _
unreachable for controls _ @xmath25 .
later on , they will be also specified as _
unobservable _ in a proper sense .
their properties is the main subject of our paper .
recall some of the known facts . for an integer @xmath149 , denote @xmath150 .
introduce a class of polynomials of variable @xmath151 @xmath152 let @xmath153 be the standard spherical harmonics on @xmath124 : @xmath154 where @xmath155 , and @xmath156 is the beltrami - laplace operator on @xmath124 . as is shown in @xcite , @xcite , the representation @xmath157 is valid ( recall that @xmath158 ) . one more characterization is @xmath159 where @xmath160 , so that @xmath161 consists of @xmath162-polyharmonic functions . note that @xmath163 holds .
* remark * a lack of local controllability in system @xmath111 is , in a sense , partially compensated by the following property .
it can be shown that for any @xmath164 one has @xmath165 i.e. , the forward parts of delayed incoming waves possess a local completeness . with the dual system one
associates an _ observation operator _
@xmath171 defined by @xmath172((s+\tau ) \omega ,- s ) , \,\,(\tau,\omega ) \in [ 0,\infty ) \times s^2\ ] ] on @xmath8-smooth rapidly decaying functions .
the duality relation is @xmath173 where @xmath174 is the radon transform : see @xcite , @xcite , @xcite . for any @xmath175 and @xmath25 , one has @xmath176 since @xmath177 . by arbitrariness of @xmath113 ,
we get @xmath178 , so that the elements of @xmath7 can be specified as _ unobservable _ ( at infinity ) in the interval @xmath179 . fix @xmath180 .
let @xmath168 in ( [ eq2 * ] ) be such that @xmath181 ( so that @xmath182 ) , @xmath81 be @xmath8-smooth in @xmath42 outside the sphere @xmath183 , and @xmath184 \,=\,\alpha(\omega ) , \quad \omega \in s^2 .
\end{aligned}\ ] ] with a nonzero @xmath99 .
thus , we have @xmath185 that yields @xmath186 where @xmath187 are the characteristic cones ( see fig 1b and compare with ( [ sing supp y ( ball ) ] ) , ( [ sing supp v^y ( ball ) ] ) ) . in particular , the cone @xmath188 supports a jump of @xmath100 and @xmath92 , whereas the geometrical optics provides @xmath189\,\\ \label{jump d / dr v^y ( mathbb rr3 ^ 3)}&=\,- \frac{\xi_0}{\xi_0-t } \alpha(\omega ) , \qquad \omega \in s^2,\,\,t<0 \end{aligned}\ ] ] that is the same as ( [ jump d / dr v^y ( ball ) ] ) .
so , the radial derivative @xmath190 has a jump at @xmath191 , which moves ( in the reversed time ) to infinity .
up to the factor @xmath192 of geometric nature , its shape reproduces the shape of the initial jump ( [ jump y ( mathbb rr3 ^ 3 ) ] ) of the cauchy data . by the use of technique @xcite ( see lemma 3 ) with regard to ( [ jump d / dr v^y ( mathbb rr3 ^ 3 ) ] ) , one can derive that the limit passage ( [ def o in mathbb rr3 ^ 3 ] ) provides the relation @xmath193\,=\,-
\xi_0 \alpha(\omega ) , \quad \omega \in s^2 , \end{aligned}\ ] ] in perfect analogy to ( [ jump o^ty ( ball ) ] ) .
thus , any jump of the cauchy data @xmath81 on a sphere propagates along the space - time rays preserving its shape , reaches the infinity , and is observed ( as a jump of @xmath194 ) in the interval @xmath195 at the moment @xmath196 .
therefore , @xmath197 and we can choose a control @xmath25 such that @xmath198 since @xmath199 and @xmath200 .
hence , one can claim definitely that @xmath201 .
in other words , no @xmath175 can have a jump on a sphere because it is visible at infinity .
as is easy to recognize , such a jump is also forbidden for any derivative @xmath202 .
surely , the embedding @xmath203 does nt mean that the singularity of @xmath81 is necessarily a jump .
however , the above - considered example motivates to suggest that elements of @xmath7 are smooth at least with respect to @xmath9 and , in particular , this embedding should be mutually exclusive with @xmath204 .
one more pro - argument is that @xmath7 is spanned on @xmath8-smooth polyharmonic functions : see ( [ d = sun dl ] ) , ( [ polyharmonic ] ) .
so , we conjectured that _ any _ spherical singularity is visible at infinity .
take @xmath205 ; let @xmath206 .
choose a ( real ) sequence @xmath207 provided @xmath208 for every @xmath209 , choose harmonics @xmath210 such that @xmath211 .
compose @xmath212^{2k+1}y_{6k+3}(\omega ) , & r\geqslant 1\ , .
\end{cases}\ ] ] the summands in ( [ h ] ) rapidly decrees as @xmath213 , so that @xmath214 is square - summable in @xmath42 and @xmath215 since @xmath216 .
moreover , as is seen from ( [ d = sun dl ] ) , one has @xmath217 . by the first property , for any fixed @xmath222 , the series in ( [ h ] ) rapidly converges in @xmath223 and @xmath224 holds .
the same can be easily checked for the derivatives : @xmath225 . as a result , we have @xmath226 .
hence , @xmath227 .
show that @xmath228 is nonempty .
take @xmath229 . by ( [ p(1/r ) ] )
, one has @xmath230 so that @xmath231 by the choice of @xmath232 and orthogonality of @xmath233 . in the mean time , applying the beltrami - laplace operator , we have @xmath234 ^ 2\,\overset{(\ref{sequence a})}=\,\infty\,.\ ] ] therefore , @xmath235 , and the element defined by ( [ h ] ) does disprove the conjecture .
note in addition that simple analysis provides for the derivatives @xmath236 also , we can make the counterexample even more expressive by taking @xmath237 . in this case
, one has @xmath238 and @xmath239 .
an open problem is to describe admissible singularities of unobservable states belonging to @xmath240 . in particular
, can one characterize the spherical singularities _
invisible at infinity _ ?
the case of a ( visible ) jump shows that one should use more subtle distribution characteristics than @xmath241 .
probably , it is a wave front . for a jump ( [ jump y ( mathbb rr3 ^ 3 ) ] ) with a positive @xmath99
, one has @xmath242=\{\gamma
\times n_\gamma\backslash 0\,|\,\,\gamma \in \gamma^{\xi_0}\}$ ] , where @xmath243 is the subspace conormal to @xmath183 @xcite .
however , to recognize , what is @xmath244 $ ] of @xmath214 defined by ( [ h ] ) , seems to be rather difficult . does @xmath244 $ ] contain conormal directions ?
the question is also open . | the paper deals with a dynamical system @xmath0 where @xmath1 is a solution ( _ wave _ ) , @xmath2 is a _ control_. for the reachable sets @xmath3 , the embedding @xmath4 holds , whereas the subspaces @xmath5 of unreachable ( _ unobservable _ ) states are nonzero for @xmath6 .
there was a conjecture motivated by some geometrical optics arguments that the elements of @xmath7 are @xmath8-smooth with respect to @xmath9 .
we provide rather unexpected counterexamples of @xmath10 with @xmath11 . |
the inner structure of hilbert spaces is easily described by fixing a basis , orthonormalizing it and working with the coordinates of every element with respect to the latter . considering finite - dimensional hilbert spaces as free @xmath0-modules one could ask whether similar generating sets of finitely generated projective c*-modules
can be indicated which characterize the modules up to isomorphism .
unfortunately , there are two obstacles : projective c*-modules need not to be free in general , and there does not exist any general notion of ` c*-linear independence ' of sets of generators because of the existence of zero - divisors in any non - trivial c*-algebra . beside these circumstances
one does not know any canonical method to replace the process of gram - schmidt or symmetric orthogonalization of bases in the situation of sets of modular generators . in 1997 , working with frames for hilbert spaces that arise canonically in wavelet and weyl - heisenberg / gabor frame theory , we got the idea to investigate modular frames of hilbert c*-modules over unital c*-algebras as a possible replacement for the questionable analogs of bases .
this class of c*-modules includes finitely generated projective ones .
the resulting theory@xcite has been encouraging because of its consistency and strength , and also because of the number of mathematical problems which can make use of it . in the field of wavelet and frame theory and
its applications to signal and image processing many frames arise as the result of group actions on single functions . extending the group to its ( reduced ) group c*-algebra or to its group von neumann algebra and taking the generated frame as a generating set of a hilbert c*-module over one of these c*-algebras we are in the context in which our concept can be applied .
this point of view has been of interest e.g. to m. a. rieffel , to o. bratteli and p. e. t. jrgensen , and to p. g. casazza and m. lammers as we know from ongoing discussions . in the literature
we found other fields of applications like the description of conditional expectations of finite ( jones ) index on c*-algebras ( @xcite ) , the analysis of cuntz - krieger - pimsner algebras ( @xcite ) , the investigation of the stable rank of c*-algebras ( @xcite ) and the search for @xmath1-invariants in global analysis ( @xcite ) .
the purpose of the present paper is to give a survey on our results on modular frames for hilbert c*-modules indicating their generality and strength as well as pointing out differences to the hilbert space frame theory and open problems . for full proofs and more details
we refer to our basic publications ( @xcite ) . the next section we explain decomposition and reconstruction results . in the third section
we deal with frame - related invariants of finitely generated projective c*-modules that characterize them up to isomorphism .
the fourth section is devoted to a structure theorem on the nature of operators @xmath2 on a certain hilbert space such that @xmath3 .
the last section is concerned with a discussion on various problems of frame approximation by ( normalized ) tight ones .
the concept of hilbert c*-modules arose as a generalization of the notions ` hilbert space ' , ` fibre bundle ' and ` ideal ' .
the basic idea has been to consider modules over arbitrary c*-algebras instead of linear spaces and to allow the inner products to take values in those c*-algebras of coefficients being c*-(anti-)linear in their arguments . for the history and for comprehensive accounts
we refer to the publications by e. c. lance@xcite , by n .- e .
wegge - olsen@xcite and by i. raeburn , d. p. williams@xcite .
let @xmath4 be a ( unital ) c*-algebra and @xmath5 be a ( left ) @xmath4-module .
suppose that the linear structures given on @xmath4 and @xmath5 are compatible , i.e. @xmath6 for every @xmath7 , @xmath8 and @xmath9 .
if there exists a mapping @xmath10 with the properties ( ) @xmath11 for every @xmath9 , @xmath12 if and only if @xmath13 , @xmath14 for every @xmath15 , @xmath16 for every @xmath8 , every @xmath17 , @xmath18 for every @xmath19 , then the pair @xmath20 is called a _ ( left ) pre - hilbert @xmath4-module_. the map @xmath21 is said to be an _ @xmath4-valued inner product_. if the pre - hilbert @xmath4-module @xmath20 is complete with respect to the norm @xmath22 then it is called a _
hilbert @xmath4-module_. two hilbert @xmath4-modules are _ unitarily isomorphic _ if there exists a bounded @xmath4-linear isomorphism of them which preserves the inner product values . in case
@xmath4 is unital the hilbert @xmath4-module @xmath5 is _ ( algebraically ) finitely generated _ if there exists a finite set @xmath23 such that @xmath24 for every @xmath9 and some coefficients @xmath25 . if @xmath4 is unital the hilbert @xmath4-module @xmath5 is _ countably generated _ if there exists a countable set @xmath26 such that the set of all finite @xmath4-linear combinations @xmath27 , @xmath25 , is norm - dense in @xmath5 .
hilbert c*-modules appear naturally in a number of situations .
for example , the set of all essentially bounded measurable maps of a measure space @xmath28 into a measurable field of hilbert spaces on @xmath28 becomes a hilbert @xmath29-module after factorization by the set of maps that are non - zero only on sets of measure zero , ( @xcite ) .
a simple algebraic construction is the algebraic tensor product @xmath30 of a c*-algebra @xmath4 and a hilbert space @xmath31 completed with respect to the norm which is derived from the @xmath4-valued inner product @xmath32 for @xmath33 , @xmath34 .
it is denoted by @xmath35 .
this kind of hilbert c*-modules is very important : every finitely generated hilbert @xmath4-module @xmath5 can be embedded into some hilbert @xmath4-module @xmath36 for finite @xmath37 as an orthogonal summand , and every countably generated hilbert @xmath4-module can be realized as an orthogonal summand of the hilbert @xmath4-module @xmath38 in such a way that its orthogonal complement is isometrically isomorphic to @xmath39 again , ( kasparov s theorem@xcite ) . to name two further examples ,
the set of all continuous sections of a certain vector bundle over a base space @xmath28 becomes naturally a finitely generated hilbert c(@xmath28)-module , and conditional expectations on c*-algebras @xmath40 turn these c*-algebras into pre - hilbert @xmath4-modules over the image c*-subalgebra @xmath41 .
( by the way , the interrelation between vector bundles and finitely generated projective c*-modules over commutative c*-algebras is a categorical equivalence similar to gelfand s theorem for c*-algebras , see serre - swan s theorem@xcite . )
the reader should be aware that the theory of hilbert c*-modules has much more exceptional examples in comparison to hilbert space theory then one can think of . e.g. the analog of the riesz representation theorem for bounded @xmath4-linear functionals on hilbert @xmath4-modules is not valid , in general .
therefore , bounded modular operators may be non - adjointable , norm - closed hilbert @xmath4-submodules can lack the property to be direct summands , the notions of topological and orthogonal direct summands are different , and more .
there are even examples of non - countably generated banach c*-modules which can be turned into hilbert c*-modules in at least two ways , however the two c*-valued inner products are not unitarily isomorphic , and so the resulting two norms give rise to non - isometrically isomorphic banach c*-modules , ( @xcite ) .
luckily , the latter oddity can not appear for countably generated hilbert c*-modules . for a concrete example consider the c*-algebra @xmath42)$]of all continuous functions on the unit interval , its ideal @xmath43)$ ] of all continuous functions vanishing at zero and the hilbert @xmath4-module @xmath44 with the standard @xmath4-valued inner product inherited from @xmath45 .
the operator @xmath46 is non - adjointable , and the hilbert @xmath4-submodule @xmath47 is a topological direct summand , but not an orthogonal one . in the light of these circumstances
the results on the existence and on the properties of modular frames for finitely or countably generated hilbert c*-modules presented below become the more remarkable .
we start with a definition of modular frames which takes an inequality in the positive cone of the c*-algebra of coefficients as its initial point .
let @xmath4 be a unital c*-algebra and @xmath48 be a finite or countable index set .
a sequence @xmath49 of elements in a hilbert @xmath4-module @xmath5 is said to be a _ frame _ if there are real constants @xmath50 such that @xmath51 for every @xmath9 .
the optimal constants ( i.e. maximal for @xmath52 and minimal for @xmath53 ) are called _
frame bounds_. the frame @xmath54 is said to be a _ tight frame _
if @xmath55 , and said to be _ normalized _ if @xmath56 .
we consider _ standard _
( normalized tight ) frames in the main for which the sum in the middle of the inequality ( [ ineq - frame ] ) always converges in norm . for non - standard frames
the sum in the middle converges only weakly for at least one element of @xmath5 .
a sequence @xmath57 is said to be a _ standard riesz basis of @xmath5 _ if it is a standard frame and a generating set with the additional property that @xmath4-linear combinations @xmath58 with coefficients @xmath59 and @xmath60 are equal to zero if and only if in particular every summand @xmath61 equals zero for @xmath62 . a generating sequence @xmath57 with the described additional property alone
is called a _ hilbert basis of @xmath5_. an _ inner summand of a standard riesz basis _ of a hilbert @xmath4-module @xmath63 is a sequence @xmath57 in a hilbert @xmath4-module @xmath5 for which there is a second sequence @xmath64 in a hilbert @xmath4-module @xmath65 such that @xmath66 and the sequence consisting of the pairwise orthogonal sums @xmath67 in the hilbert @xmath4-module @xmath68 is the initial standard riesz basis of @xmath63 .
two frames @xmath57 , @xmath64 of hilbert @xmath4-modules @xmath69 , @xmath70 , respectively , are _ unitarily equivalent _ ( resp . , _ similar _ ) if there exists a unitary ( resp . , invertible adjointable ) linear operator @xmath71 such that @xmath72 for every @xmath73 .
analyzing this definition we do not know whether a frame is a generating set , or not .
this will turn out to hold only during our investigations .
we observe that for every ( normalized tight ) frame @xmath74 of a hilbert space @xmath31 the sequence @xmath75 is a standard ( normalized tight ) module frame of the hilbert @xmath4-module @xmath76 with the same frame bounds .
so standard modular frames exist in abundance in the canonical hilbert @xmath4-modules . at the same time
wavelet theorists see that the c*-algebra @xmath4 opens up an additional degree of freedom for constructions and investigations . for the existence of standard modular frames in arbitrary finitely or countably
generated hilbert @xmath4-modules we obtained the following simple fact : ( @xcite ) for every @xmath4-linear partial isometry @xmath77 on @xmath78 ( or @xmath79 ) the image sequence @xmath80 of the standard orthonormal basis @xmath81 is a standard normalized tight frame of the image @xmath82 ( or @xmath83 ) .
consequently , every algebraically finitely generated or countably generated hilbert @xmath4-module @xmath5 possesses a standard normalized tight frame since they can be embedded into these standard hilbert @xmath4-modules as orthogonal summands .
does every hilbert c*-module admit a modular frame ?
the main property of frames for hilbert spaces is the existence of the reconstruction formula that allows a simple standard decomposition of every element of the spaces with respect to the frame .
we found that almost all the results for the hilbert space situation described in ( @xcite ) can be recovered .
sometimes the way of proving is exceptional long , for example to show that modular riesz bases @xmath74 that are normalized tight frames have to be orthogonal bases for which the values @xmath84 are all projections .
let us first formulate the reconstruction formula for normalized tight frames without the restriction to be standard : ( th .
4.1@xcite ) let @xmath4 be a unital c*-algebra , @xmath5 be a finitely or countably generated hilbert @xmath4-module and @xmath57 be a normalized tight frame of @xmath5 .
then the _ reconstruction formula _
@xmath85 holds for every @xmath9 in the sense of convergence w.r.t .
the topology that is induced by the set of semi - norms @xmath86 .
the sum converges always in norm if and only if the frame @xmath57 is standard .
conversely , a finite set or a sequence @xmath57 satisfying the formula ( [ eq1 ] ) for every @xmath9 is a normalized tight frame of @xmath5 . for a proof we have to refer to ( @xcite ) since the proof is to long to be reproduced here , and the statement for non - standard normalized tight frames is some kind of summary of the entire work done . with the experience on the possible oddities of hilbert c*-module
theory in comparison to hilbert space theory the following crucial fact is surprising because of the generality in which it holds .
the existence and the very good properties of the frame transform of standard frames give the chance to get far reaching results analogous to those in the hilbert space situation .
again , the proof is more complicated than the known one in the classical hilbert space case , cf .
( @xcite ) .
4.2@xcite ) let @xmath4 be a unital c*-algebra , @xmath5 be a finitely or countably generated hilbert @xmath4-module and @xmath57 be a standard frame of @xmath5 .
the _ frame transform of the frame @xmath57 _ is defined to be the map @xmath87 that is bounded , @xmath4-linear , adjointable and fulfills @xmath88 for a standard orthonormal basis @xmath81 of the hilbert @xmath4-module @xmath79 and all @xmath73 .
moreover , the image @xmath89 is an orthogonal summand of @xmath79 . for normalized tight frames we additionally get @xmath90 for any @xmath73 , and
@xmath91 is an isometry in that case .
the frame transform @xmath91 is the proper tool for the description of standard frames .
6.1@xcite ) let @xmath4 be a unital c*-algebra , @xmath5 be a finitely or countably generated hilbert @xmath4-module and @xmath57 be a standard frame of @xmath5 .
then the _ reconstruction formula _
@xmath92 holds for every @xmath9 in the sense of norm convergence , where the operator @xmath93 is positive and invertible .
the sequence @xmath94 is a standard frame again , the _ canonical dual frame of _
the operator @xmath95 is called the _ frame operator of @xmath57 _ on @xmath5 .
the two theorems on reconstruction above bring to light some key properties of frame sequences : ( th .
5.4 , 5.5@xcite ) every standard frame of a finitely or countably generated hilbert @xmath4-module is a set of generators .
every finite set of algebraic generators of a finitely generated hilbert @xmath4-module is a ( standard ) frame .
every standard frame can be realized as the image of a standard normalized tight frame by an adjointable invertible bounded module operator .
in particular , every standard riesz basis is the image of an orthogonal hilbert basis @xmath74 with projection - valued values @xmath96 by an adjointable invertible bounded module operator .
let us show that there exists countably generated hilbert c*-modules without standard riesz bases : let @xmath42)$ ] be the c*-algebra of all continuous functions on the unit interval @xmath97 $ ] and consider its ideal and hilbert @xmath4-submodule @xmath98)$ ] of all continuous functions vanishing at zero .
the module @xmath5 is countably generated by the stone - weierstrass theorem .
so it admits normalized tight frames .
however , if it would have a standard riesz basis , then it would contain an orthogonal hilbert basis @xmath74 with non - trivial projection - valued values @xmath99 . at the other side , the only projection contained in the range of the @xmath4-valued inner product
is the zero element .
so it does not contain any standard riesz basis .
we close our considerations on the frame transform arising from standard frames with a statement on the relation between unitary equivalence ( or similarity ) of frames and the characteristics of the image of the frame transform @xmath91 .
? * th . 7.2 ) ) let @xmath4 be a unital c*-algebra and @xmath57 and @xmath64 be standard ( normalized tight ) frames of hilbert @xmath4-modules @xmath100 and @xmath101 , respectively .
the following conditions are equivalent : ( ) the frames @xmath57 and @xmath64 are unitarily equivalent or similar
. their frame transforms @xmath102 and @xmath103 have the same range in @xmath79 .
the sums @xmath104 and @xmath105 equal zero for exactly the same banach a - submodule of sequences @xmath106 of @xmath79 .
these results look pretty much the same as for the hilbert space situation what makes them easy to apply . however , the proofs are more difficult , and they required an extensive search for possible counterexamples and obstacles beforehand .
there are also statements that do not transfer to hilbert c*-modules .
for example , standard riesz bases of hilbert c*-modules may have more than one dual frame because of the existence of zero - divisors in the c*-algebra of coefficients , see example 6.4@xcite and corollary 6.6@xcite .
more results on disjointness and inner sums of frames , as well as on various kinds of frame decompositions can be found in ( @xcite ) .
our basic publications@xcite also contain a number of illustrating examples and counterexamples we would like to refer to .
the best way for the description of the inner structure of hilbert spaces is the selection of a ( ( ortho- ) normal ) basis and the characterization of elements by their coordinates .
the notion of a basis makes essentially use of the notion of linear independence of vectors . turning to finitely generated projective c*-modules over a certain fixed unital c*-algebra @xmath4 we are most often faced with the absence of a reasonable notion of @xmath4-linear independence of sets of module elements , e.g. finite sets of algebraic generators . also , very often we have a lot of non - isomorphic @xmath4-modules possessing generating sets with the same number of algebraic generators .
the task is to find additional invariants for the distinction of projective @xmath4-modules in such situations .
fortunately , any set of algebraic generators of a finitely generated projective c*-module @xmath5 over a unital c*-algebra @xmath4 is a frame , a fact shown by the authors@xcite in 1998 .
furthermore , for every set of algebraic generators @xmath107 of @xmath5 there exists an @xmath4-valued inner product @xmath21 on @xmath5 turning this set @xmath107 into a normalized tight frame .
we show that the knowledge of the values @xmath108 turns out to be sufficient to describe the @xmath4-module @xmath5 up to uniqueness .
note that the elements @xmath109 need not to be ( @xmath4-)linearly independent , in general .
let @xmath4 be a unital c*-algebra and let @xmath110 and @xmath111 be two finitely generated hilbert @xmath4-modules .
then the following conditions are equivalent : ( ) @xmath112 and @xmath113 are isometrically isomorphic as banach @xmath4-modules .
@xmath114 and @xmath115 are unitarily isomorphic as hilbert @xmath4-modules
. there are finite normalized tight frames @xmath116 and @xmath117 of @xmath5 and @xmath65 , respectively , such that @xmath118 , @xmath119 and @xmath120 for any @xmath121 , and @xmath122 for any @xmath123 .
the equivalence of the conditions ( i ) and ( ii ) has been shown for countably generated hilbert @xmath4-modules in theorem 4.1@xcite . the implication ( ii)@xmath124(iii ) can be seen to hold setting @xmath125 for the existing unitary operator @xmath126 and for @xmath121 .
the demonstration of the inverse implication requires slightly more work . for
the given normalized tight frames @xmath109 and @xmath117 of @xmath5 and @xmath65 , respectively , we define a @xmath4-linear operator @xmath77 by the rule @xmath127 , @xmath121 .
for this operator @xmath77 we obtain the equalities @xmath128 which hold for every @xmath9 . consequently , @xmath129 for any @xmath9 , and the operator @xmath77 is unitary .
every finitely generated projective @xmath4-module @xmath5 over a unital c*-algebra @xmath4 can be reconstructed up to isomorphism from the following data : ( ) a finite set of algebraic non - zero modular generators @xmath130 of @xmath5 . a symmetric @xmath131 matrix @xmath132 of elements from @xmath4 , where @xmath133 is supposed to be equal to @xmath134 for @xmath123 and for the ( existing and unique ) @xmath4-valued inner product @xmath135 on @xmath5 that turns the set of algebraic modular generators @xmath136 into a normalized tight frame of the hilbert @xmath4-module @xmath137 .
the number of elements in sets of algebraic modular generators of @xmath5 has a minimum , and it suffices to consider sets of generators of minimal length . then the modular invariants can be easier compared permuting the elements of the generating sets if necessary .
we have already pointed out that the set of algebraic generators @xmath138 of @xmath5 is a frame with respect to any @xmath4-valued inner product on @xmath5 which turns @xmath5 into a hilbert @xmath4-module .
that means the inequality @xmath139 is satisfied for two finite positive real constants @xmath140 and any @xmath141 , see theorem 5.9@xcite .
what is more , for any frame of @xmath5 there exists another @xmath4-valued inner product @xmath142 on @xmath5 with respect to which it becomes normalized tight .
the latter inner product is unique as shown by corollary 4.3 , theorem 6.1@xcite and theorem 4.4@xcite .
so assertion ( iv ) of the previous theorem demonstrates the complete assertion .
we can say more in case the finitely generated hilbert @xmath4-module contains a modular riesz basis , i.e. a finite set of modular generators @xmath143 such that the equality @xmath144 holds for certain coefficients @xmath145 if and only if @xmath146 for any @xmath121 .
obviously , a modular riesz basis is minimal as a set of modular generators , i.e. we can not drop any of its elements preserving the generating property .
however , there can exist totally different riesz bases for the same module that consist of less elements , cf .
example 1.1@xcite .
note that the coefficients @xmath147 can be non - trivial even if @xmath146 for any index @xmath148 since every non - trivial c*-algebra @xmath4 contains zero - divisors . not every hilbert c*-module with a normalized tight modular frame does possess a modular riesz basis .
for an example we refer to example 2.4@xcite . in case of finitely
generated projective w*-modules ( and therefore , in the case of hilbertian modules over finite w*-algebras ) we are in the pleasant situation that they always contain a modular riesz basis by w. l. paschke s theorem 3.1@xcite . moreover , by spectral decomposition every element @xmath149 of a hilbert w*-module @xmath5 has a carrier projection of @xmath150 contained in the w*-algebra of coefficients @xmath4 .
so we can ascertain the following fact : let @xmath5 be a finitely generated projective @xmath4-module over a w*-algebra @xmath4 that possesses two finite modular riesz bases @xmath151 and @xmath152 .
then there exists an @xmath153 matrix @xmath154 , @xmath155 , @xmath156 , with entries from @xmath4 such that @xmath157 for any @xmath158 , and analogously , there exists a @xmath159 matrix @xmath160 with entries from @xmath4 such that @xmath161 for any @xmath162 .
suppose the left carrier projections of @xmath163 and @xmath164 equal the carrier projections of @xmath165 and @xmath166 , respectively , and the right carrier projection of @xmath163 and @xmath164 equal the carrier projections of @xmath166 and @xmath167 , respectively .
then the matrices @xmath168 and @xmath169 are moore - penrose invertible in @xmath170 and @xmath171 , respectively .
the matrix @xmath168 is the moore - penrose inverse of @xmath169 , and vice versa . since both the modular riesz bases are sets of modular generators of @xmath5 we obtain two @xmath4-valued ( rectangular , w.l.o.g . ) matrices @xmath172 and @xmath160 with @xmath155 and @xmath162 such that @xmath173 combining these two sets of equalities in both the possible ways we obtain @xmath174 for @xmath155 , @xmath162 .
now , since we deal with sets of coefficients @xmath175 and @xmath176 that are supposed to admit special carrier projections , the coefficients in front of the elements @xmath177 and @xmath178 at the right side can only take very specific values : @xmath179 where @xmath180 is the kronecker symbol , @xmath181 is the carrier projection of @xmath182 and @xmath183 is the carrier projection of @xmath184 .
so @xmath185 and @xmath186 are positive idempotent diagonal matrices with entries from @xmath4 .
the moore - penrose relations @xmath187 , @xmath188 , @xmath189 and @xmath190 turn out to be fulfilled . in total
we found a convinient way to characterize finitely generated c*-modules over unital c*-algebras up to modular isomorphism by a small amount of additional elements of the c*-algebra of coefficients derived from the set of algebraic generators and from the module structure .
one of the classical problems of operator theory is the following : given a ( finite or infinite ) sequence @xmath191 of bounded operators on a certain separable hilbert space @xmath31 fulfilling the equality @xmath192 , determine the nature of the operators @xmath191 .
a first account was found by r. v. kadison and j. r. ringrose in ii.11.2.24@xcite using dilation methods , i.e. enlarging the hilbert space @xmath31 and subsequently extending the operators .
we will use modular frame methods to resolve this problem without changing the hilbert space @xmath31 . that way we demonstrate the power of the developed methods .
some particular examples complete the picture .
let @xmath193 be a sequence of elements with the property that @xmath194 in the sense of weak convergence .
( in particular , the sequence could be norm - convergent , or only finitely many elements could be unequal to zero . ) then there exists a projection @xmath195 , @xmath196 via @xmath197 , @xmath198 , and a sequence of partial isometries @xmath199 such that : ( ) @xmath200 , @xmath201 are projections in @xmath202 similar to @xmath203 , @xmath204 for any @xmath205 , @xmath206 ; the equality @xmath207 is valid for every @xmath208 and partial isometries @xmath209 , each @xmath210 connecting the left carrier projections of @xmath211 and of @xmath212 , respectively .
let @xmath213 be the set of all linear bounded operators on the hilbert space @xmath214 , and recall that @xmath215 as a direct sum of copies of the hilbert space @xmath214 itself .
consequently , there are projections @xmath216 such that @xmath217 for any @xmath218 , @xmath219 , and @xmath220 via partial isometries @xmath221 with @xmath222 and @xmath223 .
define @xmath224 , i.e. the set of all sequences of elements of @xmath4 for which the series converges weakly in @xmath213 .
it can be identified with the set of all bounded @xmath4-linear maps @xmath225 of the hilbert @xmath4-module @xmath79 into @xmath4 setting @xmath226 , where @xmath227 . since @xmath213 is a w*-algebra , the set @xmath228 becomes a self - dual hilbert @xmath4-module ( @xcite ) .
in fact , @xmath228 is isometrically isomorphic to @xmath213 as a hilbert @xmath4-module . to see this fix
an orthonormal basis @xmath229 of @xmath228 and an orthonormal basis @xmath230 of @xmath4 and define @xmath231 @xmath232 the freedom of choice for this isomorphism is the careful selection of the orthonormal bases of the self - dual hilbert @xmath4-modules @xmath228 and @xmath4 .
if the sequence @xmath193 is given as described above then it is a ( possibly non - standard ) normalized tight frame of the self - dual hilbert @xmath4-module @xmath213 since we have @xmath233 for every @xmath234 .
there exists a frame transform @xmath235 defined by @xmath236 .
the image of @xmath91 is a direct summand and self - dual hilbert @xmath4-submodule of @xmath228 .
moreover , @xmath91 is an isometry . continuing the isometry @xmath91 to an isometry @xmath237 using the isometric isomorphism between @xmath228 and @xmath4 we obtain @xmath238 for any @xmath8 .
the structure of direct orthogonal summands of the hilbert @xmath202-module @xmath202 is well - known , they are all generated by multiplying @xmath202 by a specific orthogonal projection from the right .
so we can characterize the image of @xmath239 in @xmath202 as @xmath240 for some @xmath241 of @xmath202 .
note that @xmath242 since the module generator @xmath203 is mapped to the module generator @xmath243 .
let @xmath244 be the isometry linking @xmath243 to @xmath203 with @xmath197 , @xmath198 .
now , we establish some information on the adjoint operator of @xmath239 : @xmath245 consequently , we get the frame decomposition @xmath246 for any @xmath247 .
the frame coefficients @xmath248 may be not the optimal ones . by prop .
6.6@xcite we have the general inequality : @xmath249 for any @xmath247 . by theorem 2.1@xcite
we obtain @xmath207 for some elements @xmath250 with @xmath251 and @xmath252 of course the root of @xmath253 seems to be selected in an artificial way , the element @xmath254 would do the job as well .
however , the following inequality gives some more information on the background of the choice made : @xmath255 since the left end equals the right end and @xmath256 holds for every @xmath247 the equality @xmath257 turns out to be valid for every @xmath247 .
consequently , the linking elements @xmath250 can be selected as partial isometries of @xmath202 mapping the left carrier projection of @xmath211 to the left carrier projection of @xmath212 for each @xmath247 because @xmath202 is a von neumann algebra and any root of a given positive operator can be described this way . by the inequality ( [ eq-1 ] ) the left carrier projection of @xmath212 has to be lower - equal than the left carrier projection of @xmath211 for any @xmath247 .
suppose , the set of operators @xmath191 is a set of pairwise orthogonal ( positive ) projections @xmath216 defined on @xmath214 .
this is the simplest situation one can think of .
then @xmath258 and @xmath259 for any index @xmath148 .
furthermore , there is the classical situation of generators of cuntz algebras @xmath260 and @xmath261 , where the operators @xmath191 are partial isometries themselves , with the properties required above . here
@xmath258 and @xmath262 for any index @xmath148 .
the partial isometries @xmath210 equal to the left carrier projections of the partial isometries @xmath212 in the given situation . generally speaking , the crucial rule is played by the projection @xmath243 corresponding to the sequence @xmath191 via its frame transform , and by its partition @xmath263 by a chain of pairwise orthogonal projections @xmath216 summing up to one and each being similar to the identity operator on @xmath214 .
in the present section we consider a question on the approximation of frames of hilbert spaces @xmath31 by ( normalized ) tight ones that is related to certain methods of orthogonalization and renormalization of hilbert bases .
we have to resort to hilbert spaces instead of hilbert c*-modules since the problem is too complex to be treated in full generality .
we give comments on the more general setting whereever possible .
unfortunately , we did not find a final solution of the problem , rather we obtained hints to the complexity and difficulty of it .
the partial results are nevertheless worth to be discussed . generally speaking ,
most distance measures on sets of frame operators seem to have rather an @xmath264-character than an @xmath1-character , exept the hilbert - schmidt norm .
so the stressed for uniqueness of best approximating tight frames often can not be obtained .
are there distance measures on the set of frames of hilbert subspaces @xmath265 of @xmath31 with respect to which a multiple of the normalized tight frame @xmath266 is the closest ( normalized ) tight frame to the given frame @xmath74 of the hilbert subspace @xmath267 , or at least one of the closest ( normalized ) tight frames ?
if there are other closest ( normalized ) tight frames with respect to the selected distance measures , do they span the same hilbert subspaces of @xmath31 ? if not , how are the positions of these subspaces with respect to @xmath268 ?
let @xmath269 be an orthonormal basis of the hilbert space @xmath270 .
for the analysis operator @xmath271 , @xmath272 , of a subspace frame @xmath74 of @xmath265 there is a polar decomposition @xmath273 , where @xmath274 denotes the frame operator . we can easily check that @xmath275 for any @xmath276 .
also the projection @xmath277 plays an important role . looking through the literature
there are two approaches to this problem , one due to r. balan@xcite and the other due to v. i. paulsen , t. r. tiballi@xcite and the first author@xcite .
r. balan@xcite starts with the definition of three distance measures for pairs of frames : the frame @xmath74 of the hilbert space @xmath31 is said to be _ quadratically close _ to the frame @xmath278 of @xmath31 if there exists a non - negative number @xmath52 such that the inequality @xmath279 is satisfied .
the infimum of all such constants @xmath52 is denoted by @xmath280 . in general , if @xmath281 then @xmath282 , however this distance measure is not reflexive .
two frames @xmath74 and @xmath278 of a hilbert space @xmath31 are said to be _ near _ if @xmath283 .
they are near if and only if they are similar , ( th .
2.4@xcite ) .
the distance measure @xmath284 is an equivalence relation and fulfils the triangle inequality .
3.1@xcite ) for a given frame @xmath74 of @xmath31 with frame bounds @xmath140 the distance measures admit their infima at @xmath285 these values are achieved by the tight frames @xmath286{cd } s^{1/2}(x_i ) \}_i \ , , \ ] ] listed in the same order as the measures above .
the solution may not be unique , in general , however any tight frame @xmath278 of @xmath31 that achieves the minimum of one of the three distance measures @xmath280 , @xmath287 and @xmath284 is unitarily equivalent to the corresponding solutions listed above .
the difference of the connecting unitary operator and the product of minimal distance times either @xmath288 or @xmath289 fulfils a certain measure - specific operator norm equality which can be found at th .
we point out that the first constant at ( [ eqn1 ] ) is the arithmetic mean of @xmath290 and @xmath291 , the second one is their harmonic mean and the third one is their geometrical mean .
2.3@xcite ) the operator @xmath292 is hilbert - schmidt if and only if the sum @xmath293 is finite for at least one normalized tight frame @xmath294 of a hilbert subspace @xmath295 of @xmath31 that is similar to @xmath74 . in this situation
the estimate @xmath296 is valid for every normalized tight frame @xmath294 of any hilbert subspace @xmath295 of @xmath31 that is similar to @xmath74 , where @xmath297 denotes the hilbert - schmidt norm .
( the left sum can be infinite for some choices of subspaces @xmath295 and normalized tight frames @xmath294 for them . )
equality appears if and only if @xmath298 for any @xmath299 .
consequently , _ the symmetric approximation of a frame @xmath74 in a hilbert space @xmath267 _ is the normalized tight frame @xmath300 spanning the same hilbert subspace @xmath301 of @xmath31 and being similar to @xmath74 via the invertible operator @xmath289 .
applying the theorem to appropriate riesz bases @xmath74 the normalized tight frame @xmath300 turns out to be the symmetric orthogonalization of this basis as discovered by p .- o .
lwdin@xcite in 1948 .
this is why the denotation ` symmetric approximation ' has been selected for the normalized tight frame @xmath300 .
the resultsin ( @xcite ) generalize to modular frames of countably generated hilbert c*-modules over commutative c*-algebras , whereas for respective hilbert c*-modules over non - commutative c*-algebras the proofs can not reproduced at several crucial places where commutativity of the c*-algebra of coefficients is essential .
so the formulation of propositions on the non - commutative case is an open problem at present . to overcome the difficulties with the non - commutativity of the c*-algebra of coefficients we consider distance - measures based on the various frame operators .
the properties of these operators do not depend on the choice of the set of coefficients ( @xcite ) .
one idea could be to consider the distance with respect to the operator norm of the difference of the orthogonal projections @xmath302 onto the ranges of the frame transforms of two given frames @xmath74 and @xmath278 .
unfortunately , two frames of a certain hilbert space @xmath31 are similar if and only if these projections coincide ( th . 7.2@xcite ) .
so we would only characterize classes of similar frames .
the better idea is to consider the operator norm distance of the respective frame transforms @xmath303 , @xmath304 or of the respective frame operators @xmath305 , @xmath306 .
the latter act as positive diagonalizable operators on the given hilbert c*-module , whereas the former map it to the standard countably generated hilbert c*-module @xmath79 .
for any tight frame @xmath278 of a hilbert space @xmath31 the corresponding frame operator @xmath306 equals to the identity operator times the frame bound value . for a given frame @xmath74 of @xmath31 with frame bounds
@xmath140 the closest positive multiple of the identity operator to the frame operator @xmath305 is @xmath307 .
unfortunately , every tight frame @xmath278 of @xmath31 with frame bound @xmath308 fulfils this condition , and the relative position of tight frames in @xmath31 is of greater importance than one can express that way .
however , we got a hint for the kind of factor to be used
. now , let us consider the norm of differences of frame transforms .
let @xmath74 be a frame of a certain hilbert space @xmath31 with frame bounds @xmath140 , and let @xmath95 be its frame operator .
the frame transform @xmath309 of the tight frame @xmath310 is the closest one w.r.t .
the operator norm to the frame transform @xmath303 of the given frame @xmath74 among all the frame transforms of positive multiples of the normalized tight frame @xmath300 .
denote the frame transform of the tight frame @xmath311 by @xmath312 for @xmath313 .
because the frame transforms @xmath312 and @xmath309 all possess the same coisometries in their respective polar decompositions we obtain the equality @xmath314 . since the inequality @xmath315 is valid and both @xmath316 and @xmath289 are diagonalizable with a common set of eigenvectors that forms a basis of @xmath31 we obtain @xmath317 the right expression is minimal if and only if @xmath318 is the arithmetic mean of the lower and the upper spectral bound @xmath290 and @xmath291 of the positive invertible operator @xmath289
. suppose the frame @xmath74 of a certain hilbert space @xmath31 is fixed and possesses the frame bounds @xmath52 and @xmath53 .
let us consider tight frames @xmath319 of @xmath31 with frame bound @xmath320 for which the norm of the difference of their frame transform @xmath304 and of the frame transform @xmath309 of the tight frame @xmath321 is small .
the next example shows that there are usually a lot of quite different tight frames of @xmath31 with frame bound @xmath320 the frame transforms of which realize the same norm distance to the frame transform @xmath303 of the initial frame as the distinguished tight frame @xmath322 of @xmath31 .
we want to point out that the example works already for finite - dimensional hilbert spaces and for frames with finitely many elements .
let @xmath323 be a separable hilbert space and fix its standard orthonormal basis @xmath269 .
set @xmath324 , @xmath325 and @xmath326 for @xmath327 .
the resulting set @xmath74 is a riesz basis of @xmath31 with frame bounds @xmath328 and @xmath329 . since @xmath330 we can consider other tight frames @xmath278 of @xmath31 with the same frame bound @xmath331 that are defined as @xmath332 for @xmath333 and @xmath334 for some @xmath335 . obviously , @xmath336 for @xmath337 and @xmath338 , where @xmath339 maps @xmath340 to @xmath341 , @xmath342 and @xmath77 is the identity map . since both @xmath289 and @xmath339
are normal , diagonalizable and commuting , with a basis consisting of common eigenvectors @xmath269 , we can estimate @xmath343 see e. a. azoff and c. davis@xcite , or k. r. davidson@xcite . the eigenvalues can be counted , they are @xmath344 and @xmath345 for @xmath333 and @xmath346 .
taking the concrete values from the definitions of both the frames we obtain that the maximum of the difference of the corresponding eigenvalues is determined by the first two terms as long as @xmath347 .
so all these tight frames @xmath278 parametrized by @xmath348 realize the same norm @xmath349 for the difference of the respective frame transforms . summarizing
, the measure of nearness of a frame to some tight frame derived from the norm of the difference of their frame transforms in general gives an entire manifold of tight frames that are ` closest ' to a given frame with respect to this measure .
moreover , if we allow the tight frame to span a probably smaller hilbert space than the original frame @xmath74 , then the condition @xmath350 to the frame bounds @xmath140 of @xmath74 turns out to be occasionally essential to guarantee that the closest tight frame spans exactly the same hilbert space than the initial frame . in other words ,
the distance between the square root of the lower frame bound @xmath52 and the arithmetic mean of the square roots of the lower and the upper frame bounds @xmath52 and @xmath53 , respectively , has to be smaller than the distance of the square root of @xmath52 to zero . the problem stated in the beginning of the present section remains unsolved despite of the encouraging partial results indicated above , even for the approximation of frames of hilbert spaces by ( normalized ) tight ones .
we will continue our work to find a solution for it .
we are grateful to radu balan , peter g. casazza , mark lammers , christian le merdy and vern i. paulsen for fruitful discussions and useful remarks on the subjects explained in the present survey , especially on the use of the modular point of view in classical frame theory .
99 , `` frames in hilbert c*-modules and c*-algebras '' , preprint , university of houston , houston , and texas a&m university , college station , texas , u.s.a . , 1998 . , `` a module frame concept for hilbert c*-modules '' , in : _ functional and harmonic analysis of wavelets ( san antonio , tx , jan .
1999 ) _ , d. r. larson , l. w. baggett , eds . , ams , providence , r.i . , _ contemp .
_ * 247 * , 207 - 233 , 2000 .
, `` index for c*-subalgebras '' , _ memoirs amer . math .
_ * 424 * , 1990 , 83 pp . ,
`` entropy and index for subfactors '' , _ ann .
scient . ec .
_ * 19 * , 57 - 106 , 1986 . , `` indice dune esperance conditionelle '' , _ compos
_ * 66 * , 199 - 236 , 1988 .
, `` on conditional expectations of finite index '' , _
j. oper . theory _
* 40 * , 87 - 111 , 1998 .
, `` the c*-algebra of a hilbert bimodule '' , _ bollettino unione mat .
b artic .
ric . mat .
( 8) _ * 1 * , no . 2 , 263 - 282 , 1998 . , `` ideal structure and simplicity of the c*-algebras generated by hilbert bimodules '' , _ j. funct .
* 159 * , 295 - 322 , 1998 .
, `` the stable rank of topological algebras and a problem of r. g. swan '' , _ j. funct .
* 160 * , 42 - 78 , 1998 .
, `` on the stable rank of simple c*-algebras '' , _ j. amer .
* 12 * , 1091 - 1102 , 1999 .
, `` @xmath1-torsion invariants '' , _ j. funct
. anal . _ * 110 * , 377 - 409 , 1992 .
, `` hilbert modules and modules over finite von neumann algebras , and applications to @xmath1-invariants '' , _ math . ann . _ * 309 * , 247 - 285 , 1997 .
, `` hilbertian versus hilbert w*-modules , @xmath1- and other invariants '' , preprint no . 9/2000 , zhs - ntz , universitt leipzig / math.oa/0003185 ( in xxx.lanl.gov ) , 2000 .
, _ hilbert c*-modules a toolkit for operator algebraists _ , london mathematical society lecture note series v. 210 , cambridge university press , cambridge , england , 1995 . , _ k - theory and c*-algebras a friendly approach _ , oxford university press , oxford , england , 1993 . , _ morita equivalence and continuous trace c*-algebras _ , mathematical surveys and monographs , v. 60 , amer
. math .
soc . , providence , r.i .
, `` direct integrals and hilbert w*-modules ( russ./engl . ) '' , in : problems in algebra , geometry and discrete mathematics , eds . :
o. b. lupanov , a. i. kostrikin , moscow state university , dept .
moscow , russia , 1992 , 162 - 177 / e - print funct - an/9312002 at xxx.lanl.gov .
, `` geometrical aspects of hilbert c*-modules '' , _ positivity _ * 3 * , 215 - 243 , 1999 . , `` frames for banach spaces '' , in : _ functional and harmonic analysis of wavelets ( san antonio , tx , jan .
1999 ) _ , d. r. larson , l. w. baggett , eds . , ams , providence , r.i . , _ contemp . math . _ * 247 * , 149 - 182 , 2000 . , `` frames , bases and group representations '' , _ memoirs amer . math .
soc . _ , 1998 , to appear .
, `` inner product modules over b*-algebras '' , _ trans .
soc . _ * 182 * , pp .
443 - 468 , 1973 . , _ fundamentals of the theory of operator algebras , i - iv _ , academic press , new york , 1983 . ,
`` factorization in c*-algebras '' , _ expos .
_ * 16 * , pp . 145 - 156 , 1983 . , `` equivalence relations and distances between hilbert frames '' , _ proc .
soc . _ * 127 * , 2353 - 2366 , 1999 . ,
`` symmetric orthogonalization of vectors in hilbert spaces '' , _ ph.d . thesis _ , university of houston , houston , texas , usa , 1991 . , `` symmetric approximation of frames and bases in hilbert spaces '' , _ trans .
_ , to appear . , `` an open problem '' , concentration week , texas a&m university , college station , tx , july 14 - 19 , 1999 , contribution . ,
`` on the nonorthogonality problem '' , _ adv .
quantum chem .
_ * 5 * , 185 - 199 , 1970 . ,
`` on distances between unitary orbits of self - adjoint operators '' , _ acta sci .
_ * 47 * , 419 - 439 , 1984 . ,
`` the distance between unitary orbits of normal operators '' , _ acta sci . math . _ * 50 * , 213 - 223 , 1986 .
, `` estimating the distance between unitary orbits '' , _ j. oper . theory _ * 20 * , 21 - 40 , 1988 . | we give a comprehensive introduction to a general modular frame construction in hilbert c*-modules and to related linear operators on them .
the hilbert space situation appears as a special case .
the reported investigations rely on the idea of geometric dilation to standard hilbert c*-modules over unital c*-algebras that admit an orthonormal modular riesz basis .
interrelations and applications to classical frame theory are indicated .
resorting to frames in hilbert spaces we discuss some measures for pairs of frames to be close to one another .
in particular , the existence and uniqueness of the closest ( normalized ) tight frame to a given frame is investigated . for riesz bases with certain restrictions
the set of closest tight frames often contains a multiple of its symmetric orthogonalization . |
disordered face - centered - cubic ( fcc ) fe@xmath7pt@xmath8 and fe@xmath9ni@xmath10 alloys have remained at the forefront of condensed matter theory for more than sixty years , owing to their rich variety of intriguing physical properties .
their linear thermal expansion coefficient ( ltec ) , @xmath11 , is anomalously small @xmath12 $ ] over a wide range of temperature @xcite , a phenomenon known as the invar effect .
their spontaneous volume magnetostriction , @xmath13 , measured at @xmath14 greatly exceeds that in body - centered - cubic ( bcc ) fe and fcc ni @xcite .
their reduced magnetostriction , @xmath15 , scales with the square of the reduced magnetization , @xmath16^{2}$ ] , up to a temperature near the curie temperature , @xmath17 @xcite .
surprisingly , only one of these two ferromagnets , namely fe@xmath9ni@xmath10 , shows a peculiar thermal dependence of the reduced magnetization @xcite . understanding all of the abovementioned phenomena within one framework is still a major open challenge .
the most common theoretical explanation for the invar effect involves the so - called 2@xmath18-state model , where the iron atoms can switch between two magnetic states with different atomic volumes as the temperature is raised @xcite .
this theory , however , seems incompatible with the results of mssbauer @xcite and neutron experiments @xcite .
another popular explanation emphasizes the importance of non - collinearity of the local magnetic moments on iron sites @xcite , though experiments undertaken to detect such non - collinearity have not found it @xcite .
an alternative scenario with a purely magnetic origin for the invar effect has been proposed @xcite : the phenomenon is caused by anomalous thermal evolution of the magnitude of fe moments .
it is supported by a recent work on iron - platinum alloys @xcite which involves _ ab initio _ density functional theory ( dft ) calculations and the disordered local moment ( dlm ) model @xcite .
however , the method employed in @xcite can not be extended to iron - nickel alloys .
thus , it is unable to provide a unified picture for the invar effect in fe@xmath7pt@xmath8 and fe@xmath9ni@xmath10 and another treatment is called for .
a theoretical framework @xcite has recently been designed to address the spontaneous magnetization , the spontaneous volume magnetostriction , and their relationship in fe@xmath7pt@xmath8 and fe@xmath9ni@xmath10 in the temperature interval @xmath19 . taking a similar approach as in @xcite and @xcite , alloys in equilibrium at temperature @xmath20
have been modelled by random substitutional alloys in homogeneous ferromagnetic ( fm ) states , partially disordered local moment ( pdlm ) states , or dlm states depending on the fraction of fe moments which are antiferromagnetically aligned with the magnetization at @xmath20 , @xmath21 .
the procedure could be divided into the following three stages . in the first stage , physical properties of interest ( volume and magnetization )
have been calculated for fm ( @xmath22 ) , pdlm ( @xmath23 ) , and dlm ( @xmath24 ) states using _ ab initio _ dft . in the second stage
, the thermal evolution of the fraction of fe moments which point down has been determined by noticing that an accurate description of the reduced magnetization is provided by a function of this form @xmath25^{q}\end{aligned}\ ] ] and assuming that @xmath5 obeys the following equation @xmath26 \bigg [ 1- \left ( \frac{t}{t_{\rm c } } \right)^{p } \bigg]^{q}.\end{aligned}\ ] ] in the third and final step , the outputs from the previous steps have been combined to explore how the magnetization and the magnetostriction vary as the system is heated . direct comparison between simulations results and experimental measurements
has provided validation for the approach .
the study supports the following ideas .
the alloys at @xmath14 share several physical properties : the magnetization in a pdlm state collapses as the fraction of fe moments which point down increases , following closely @xmath27 while the volume shrinks , following closely @xmath28 x^{{\rm fe}\downarrow } ( 1-x^{{\rm fe}\downarrow});\end{aligned}\ ] ] the volume in the fm state greatly exceeds that in the dlm state ; @xmath29 is close to 0 .
these common properties can account for a variety of intriguing phenomena displayed by both alloys , including the anomaly in the magnetostriction at @xmath14 and , more surprisingly perhaps , the scaling between the reduced magnetostriction and the reduced magnetization squared below the curie temperature .
however , the thermal evolution of the fraction of fe moments which point down depends strongly on the alloy under consideration .
this , in turn , can explain the observed marked difference in the temperature dependence of the reduced magnetization between the two alloys .
@*4c & volume ( @xmath30 ) & bulk modulus ( gpa ) & grneisen constant + fe@xmath7pt@xmath8 & 13.44 & 177 & 2 + fe@xmath9ni@xmath10 & 11.59 & 177 & 2 + fe@xmath31ni@xmath32 & 11.13 & 193 & 2 + this paper deals with the invar effect in collinear ferromagnets fe@xmath0 ( @xmath1 ) with disordered fcc structure . the rich variety of thermal expansion displayed by these materials has firmly been established by experiments @xcite .
this makes them particularly attractive for testing our general approach , identifying conditions under which an alloy shows the invar effect , and investigating the mechanism of the invar phenomenon . in principle
, the ltec can be derived from the configuration - averaged free energy which depends explicitly on volume and temperature . in practice ,
application of dft to _ ab initio _ calculations of a finite - temperature average free energy remains difficult , even in the adiabatic approximation where the electronic , the vibrational , and the magnetic contributions are treated separately .
one of the major issues in implementing this strategy is how to incorporate magnetism correctly within the current approximations to the exchange and correlation functional @xcite .
our simulation technique can be viewed as an extension of @xcite in which the vibrational contribution to the average free energy is treated within the debye - grneisen model @xcite .
section [ comp_methods ] is devoted to computational details .
section [ results ] presents a comprehensive discussion of our results .
as we shall see , this work challenges the conventional picture of the invar effect as resulting from peculiar magnetic behaviour @xcite .
$ ] [ panel ( a ) ] , the difference in bulk modulus @xmath33 $ ] [ panel ( b ) ] , and the difference in grneisen constant @xmath34 $ ] [ panel ( c ) ] plotted against the fraction of fe moments which point down for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 .
symbols show results of emto calculations .
note that the values for @xmath35 , @xmath36 , and @xmath37 are displayed in table [ table1].,width=302 ] , @xmath38 , and @xmath39 $ ] plotted against the reduced temperature @xmath40 for fe@xmath7pt@xmath8 [ panel ( a ) ] , fe@xmath9ni@xmath10 [ panel ( b ) ] , and fe@xmath31ni@xmath32 [ panel ( c)].,width=302 ]
to address the invar effect in collinear ferromagnets fe@xmath0 ( @xmath1 ) with disordered fcc structure , we extend the scheme developed in @xcite to include atomic vibrations .
fe@xmath0 alloys in equilibrium at temperature @xmath20 in the range @xmath19 are modelled by random substitutional alloys in fm , pdlm , or dlm states depending on @xmath21 .
the method remains divided into three main stages . as a first step , we perform calculations of the volume @xmath4 for various temperatures and fm ( @xmath22 ) , pdlm ( @xmath23 ) , and dlm ( @xmath24 ) states . for a fixed value of @xmath20 and @xmath5 ,
the computational process is as follows : 1 .
we calculate the total energy @xmath41 for various wigner - seitz radii .
this is done within the framework of the exact muffin - tin orbitals ( emto ) theory in combination with the full charge density ( fcd ) technique @xcite .
further details can be found in @xcite .
we deduce from the results of step ( i ) the wigner - seitz radius @xmath42 , the volume @xmath43 , the bulk modulus @xmath44 , and the grneisen constant @xmath45 @xcite .
3 . for each wigner - seitz radius chosen in step ( i )
, we estimate the contribution to the helmholtz free energy @xmath46 from the outputs of step ( ii ) @xmath47 where the vibrational energy and the vibrational entropy take the simple form @xmath48\end{aligned}\ ] ] and @xmath49 - 3 k_{\rm b } \ln [ 1-e^{-\theta ( r , x^{{\rm fe}\downarrow})/t}].\end{aligned}\ ] ] here , @xmath50 denotes the debye function . in analogy with @xcite
, we choose the debye temperature @xmath51 to be given by @xmath52^{3 \gamma ( x^{{\rm fe}\downarrow})},\end{aligned}\ ] ] where @xmath53 scales with @xmath54^{1/2}$ ] .
we take the proportionality factor from @xcite .
4 . we minimize the sum @xmath55 with respect to @xmath56 to obtain the volume @xmath4 . as a second step ,
we investigate how heating the alloy affects its fraction of fe moments which point down .
the adopted method has already been described elsewhere @xcite . in the third and final step ,
we combine the outputs from the two previous stages to explore how the volume @xmath57 $ ] and the anomalous contribution to the ltec @xmath58 vary as the temperature is raised . to allow for direct comparison between simulations and experiments @xcite , we conveniently define @xmath58 as the difference between @xmath59 and @xmath60 , where the normal contribution to the ltec measures the expansion that would occur if we heated the alloy in a dlm ( ` paramagnetic ' ) state @xmath61 ( t , 1/2).\end{aligned}\ ] ] it is instructive to reexpress @xmath58 as the sum of two terms @xmath62 [ t , x^{{\rm fe}\downarrow}(t ) ] - \bigg [ \frac{1}{3 v } \bigg ( \frac{\partial v}{\partial t } \bigg)_{x^{{\rm fe}\downarrow } } \bigg ] ( t , 1/2 ) \end{aligned}\ ] ] and @xmath63 [ t , x^{{\rm fe}\downarrow}(t ) ] \frac{dx^{{\rm fe}\downarrow}}{dt}(t)\end{aligned}\ ] ] that corresponds to two distinct sources of anomaly : one associated with the expansion that would occur if we heated the alloy _ without changing the configuration of fe moments _ and another one linked with the expansion that would occur if we changed the configuration of fe moments , _ but did not otherwise heat the system_. this latter contribution to @xmath58 can be conveniently written as the product of the prefactor @xmath64 , the magnetostructural coupling @xmath65 = \bigg [ - \frac{1}{v } \bigg ( \frac{\partial v}{\partial x^{{\rm fe}\downarrow}}\bigg)_{t}\bigg ] [ t , x^{{\rm fe}\downarrow}(t)],\end{aligned}\ ] ] and the rate at which the fraction of fe moments which point down fluctuates as the system is heated @xmath66 .
according to experiments @xcite , fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 exhibit a wide variety of thermal behaviour , the fe - rich alloys showing the invar effect and the fe - poor alloy presenting thermal expansion similar to that of a paramagnetic compound .
for this reason , they represent a suitable choice for testing the predictive power of the method developed in section [ comp_methods ] , formulating conditions for the occurrence of the invar effect , and investigating the mechanism of the phenomenon . plotted against the reduced temperature for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 .
panel ( b ) : the renormalized anomalous contribution for the two fe - rich alloys . filled symbols show results of numerical calculations .
open symbols display experimental data @xcite.,width=302 ] table [ table1 ] shows the calculated volumes @xmath35 , bulk moduli @xmath36 , and grneisen constants @xmath37 .
figure [ figure1 ] displays the calculated differences in volumes @xmath67 $ ] , bulk moduli @xmath33 $ ] , and grneisen constants @xmath34 $ ] for fm , pdlm , and dlm states .
note that the structural data have already been discussed @xcite .
regardless of the chemical nature of the alloy , the volume @xmath68 shrinks with increasing the fraction of fe moments which point down , following closely ( [ eqn3 ] ) .
the volume for the fm state and the volume for the dlm state differ by more than @xmath69 in the fe - rich alloys .
the volume difference drops to @xmath70 when switching to the fe - poor alloy .
we now turn to describe the materials response to uniform compression . whether we consider fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , or fe@xmath31ni@xmath32
, the bulk modulus for the fm state lies within 175 and @xmath71 .
this is consistent with measurements performed on fe@xmath7pt@xmath8 and ni @xcite .
the effect of raising @xmath5 on the bulk modulus @xmath72 mirrors to a certain extent that seen in panel ( a ) for the volume @xmath68 : ( i ) the bulk modulus decreases in the invar alloys , revealing that these materials become easier to squeeze .
( ii ) the difference @xmath73 $ ] , which amounts to @xmath74 in fe@xmath7pt@xmath8 , @xmath75 in fe@xmath9ni@xmath10 , and @xmath76 in fe@xmath31ni@xmath32 , is considerably larger in the fe - rich alloys .
we note in passing that these findings might shed light on anomalies observed in measurements of bulk moduli @xcite .
while we discuss figure [ figure1 ] , we point out that numerical noise poses a significant problem for the determination of the grneisen constants .
figure [ figure2 ] illustrates how the volumes @xmath77 , @xmath38 , and @xmath39 $ ] change with varying the temperature in the range @xmath19 .
a useful way to analyze these data is as follows .
imagine that the magnetic configuration were fixed ( @xmath78 ) .
let us call the corresponding curve @xmath79 $ ] ; the curve for fe@xmath7pt@xmath8 and fe@xmath31ni@xmath32 is the uppermost black curve in panels ( a ) and ( c ) .
then the material would not exhibit the invar effect .
this would also be the case if all of the curves @xmath80 for @xmath81 superimposed @xmath82 $ ] .
in reality , however , raising the temperature from @xmath83 to @xmath84 causes the material to demagnetize , and the value of @xmath5 changes accordingly .
one may say that the system hops from the curve @xmath85 $ ] to the curve @xmath86 $ ] , resulting in a volume given by the curve @xmath57 $ ] .
this is shown as a dashed line .
insofar as panel ( b ) allows us to judge for fe@xmath9ni@xmath10 , each hop is to a curve lower than the last , cancelling the upward trend of each individual curve : this is the essence of the invar effect . in section [ resultsb ] ,
we present a necessary condition under which an alloy shows the invar effect .
consistent with the analysis of figure [ figure2 ] , the criterion involves @xmath87 . in panel ( a ) of figure [ figure3 ]
, we plot the calculated anomalous contribution to the ltec @xmath58 against the reduced temperature for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 .
irrespective of the material under consideration , @xmath58 exhibits a negative sign opposite to @xmath60 .
however , only the fe - rich materials possess the exceptional property that @xmath58 compensates for @xmath60 in a wide temperature range .
thus the approach predicts the occurrence of the invar effect in fe@xmath7pt@xmath8 and fe@xmath9ni@xmath10 and its absence in fe@xmath31ni@xmath32 .
this perfectly matches experimental findings @xcite .
to further evaluate the predictive power of the method , we compare the calculated renormalized anomalous contribution to the ltec @xmath88 with experimental observations @xcite for the invar alloys in panel ( b ) of figure [ figure3 ] .
note that we extract the calculated values for @xmath89 - v(0,1/2 ) \ } / v(0,1/2)$ ] from figure [ figure2 ] and obtain 2.29% for fe@xmath7pt@xmath8 and 3.03% for fe@xmath9ni@xmath10 .
panel ( b ) of figure [ figure3 ] reveals a good quantitative agreement between simulations and experiments .
for instance , the curve for fe@xmath7pt@xmath8 intersects that for fe@xmath9ni@xmath10 at @xmath90 and 0.6 according to simulations and @xmath91 and 0.55 according to experiments .
another example involves the difference between @xmath92 of the former alloy and that of the latter estimated at @xmath93 : the calculated quantity is @xmath94 , while the corresponding measured value amounts to @xmath95 .
figure [ figure3 ] provides strong evidence that the approach presented in this paper captures the essential physics of the invar effect .
this opens exciting opportunities for identifying conditions under which an alloy shows the invar effect and investigating the mechanism of the phenomenon , which , in principle , can now be understood within the same framework as other intriguing observations @xcite , including : ( i ) the anomalously large magnetostriction in fe@xmath7pt@xmath8 and fe@xmath9ni@xmath10 at @xmath14 , ( ii ) the peculiar temperature dependence of the reduced magnetization in fe@xmath9ni@xmath10 , and ( iii ) the scaling of the reduced magnetostriction with the square of the reduced magnetization in fe@xmath7pt@xmath8 and fe@xmath9ni@xmath10 below the curie temperature .
[ panel ( a ) ] and @xmath96 [ panel ( b ) ] to @xmath58 plotted against the reduced temperature for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 .
symbols show results of numerical calculations .
hatched symbols correspond to simulations performed for the two fe - rich alloys with their magnetostructural coupling @xmath97 substituted by that of fe@xmath31ni@xmath32.,width=302 ] the decomposition of the anomalous contribution to the ltec @xmath58 into its two parts @xmath98 and @xmath96 is plotted against @xmath40 in figure [ figure4 ] for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 .
the two competing terms @xmath99(t,1/2)$ ] and @xmath99[t , x^{{\rm fe}\downarrow}(t)]$ ] balance each other almost completely , resulting in a very small @xmath100 ( i.e. , @xmath100 of the order of @xmath101 , or less ) .
it is clear that any strong deviation from zero shown by the anomalous contribution to the ltec arises from @xmath102\,dx^{{\rm fe}\downarrow}/dt(t)$ ] .
features in the structural behaviour of the materials which have been observed experimentally ( see figure [ figure3 ] ) , but have remained unexplained , can now be interpreted on the basis of the abovementioned insight and our theoretical results displayed in figure [ figure4 ] : ( i ) the drop in the anomalous contribution to the ltec in fe@xmath103ni@xmath104 at @xmath105 when the nickel concentration is reduced from 0.8 to 0.35 arises from the steep decrease of the product of the magnetostructural coupling @xmath106 $ ] and the magnetic term @xmath66 .
( ii ) the fact that the anomalous contribution to the ltec in fe@xmath7pt@xmath8 diminishes significantly as @xmath40 is raised from 0.5 to 0.9 , whereas that in fe@xmath9ni@xmath10 does not reflects the different behaviours of @xmath107 in this interval : this physical quantity decreases drastically in the fe - pt case , but remains almost constant in that of fe - ni . on the basis of figures [ figure3 ] and [ figure4 ]
, we argue that _ the invar phenomenon occurs only when the thermal expansion arising from the temperature dependence of the fraction of fe moments which point down @xmath108 compensates for the thermal expansion associated with the anharmonicity of lattice vibrations @xmath109 in a wide temperature interval_. a natural question to ask is : why do some alloys fulfill this necessary condition for the occurrence of the invar effect and others do not ? to shed light on this matter , consider our results presented in figures [ figure4 ] and [ figure5 ] . in fe@xmath31ni@xmath32 ,
the magnetostructural coupling is weak at @xmath14 ( @xmath110=0.74\,10^{-2}$ ] ) and @xmath108 fails to counterbalance @xmath109 over a broad temperature range . in the fe - rich alloys , however , the magnetostructural coupling is especially strong ( @xmath110 > 9\,10^{-2}$ ] ) and @xmath108 compensates for @xmath109 in a wide temperature interval . interestingly ,
if we substitute their magnetostructural coupling @xmath97 by that of fe@xmath31ni@xmath32 , the physical situation changes drastically , resembling that in fe@xmath31ni@xmath32 .
this supports the idea that only alloys with strong magnetostructural coupling at @xmath14 can show the invar effect .
[ panel ( a ) ] and the magnetostructural coupling @xmath106 $ ] [ panel ( b ) ] plotted against the reduced temperature for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32.,width=302 ]
to address the invar effect in collinear ferromagnets fe@xmath0 ( @xmath1 ) with disordered fcc structure , we have extended the scheme developed in @xcite to include atomic vibrations .
fe@xmath0 alloys in equilibrium at temperature @xmath20 in the range @xmath19 have been modelled by random substitutional alloys in fm , pdlm , or dlm states depending on @xmath21 .
the method has been divided into three main stages . as a first step ,
we have performed calculations of the volume @xmath4 for various temperatures and fm , pdlm , and dlm states . as a second step ,
we have investigated how heating the alloy affects its fraction of fe moments which point down . in the third and final step ,
we have combined the outputs from the two previous stages to explore how the volume @xmath57 $ ] and the anomalous contribution to the ltec @xmath58 vary as the temperature is raised .
it is worth emphasizing that neither partial chemical order @xcite nor static ionic displacement @xcite has been explicitly taken into account at any stage .
tests results for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 have provided evidence that the methodology captures the essential physics of the invar effect .
this opens exciting opportunities for investigating the mechanism of the phenomenon , which , in principle , can now be understood within the same framework as other intriguing observations @xcite .
we have decomposed the anomalous contribution to the ltec @xmath111 into two parts and studied each of them separately , for fe@xmath7pt@xmath8 , fe@xmath9ni@xmath10 , and fe@xmath31ni@xmath32 .
our results support the following criterion : the invar phenomenon occurs only when the thermal expansion arising from the temperature dependence of the fraction of fe moments which point down @xmath108 compensates for the thermal expansion associated with the anharmonicity of lattice vibrations @xmath109 in a wide temperature interval . finally , based on the study of @xmath108 and @xmath97 , we have predicted that only alloys with strong magnetostructural coupling at @xmath14 can show the invar effect .
this work challenges the conventional picture of the invar effect as resulting from peculiar magnetic behaviour .
the author thanks i. a. abrikosov ( linkping ) , b. alling ( linkping ) , c. a. hooley ( st andrews , u.k . ) , a. e. kissavos ( linkping ) , and j. neugebauer ( dsseldorf ) for fruitful discussions . | we present a necessary condition under which a collinear ferromagnet fe@xmath0 ( @xmath1 ) with disordered face - centered - cubic structure exhibits the invar effect .
the condition involves the rate at which the fraction of fe moments that are antiferromagnetically aligned with the magnetization fluctuates as the system is heated , @xmath2 .
another contributing factor is the magnetostructural coupling @xmath3 , where the volume @xmath4 corresponds to a homogeneous ferromagnetic state , a partially disordered local moment state , or a disordered local moment state depending on the value of @xmath5 . according to the criterion
, the invar phenomenon occurs only when the thermal expansion arising from the temperature dependence of the fraction of fe moments which point down @xmath6 compensates for the thermal expansion associated with the anharmonicity of lattice vibrations in a wide temperature interval . upon further investigation
, we provide evidence that only alloys with strong magnetostructural coupling at zero kelvin can show the invar effect . |
the paper is devoted to the analysis of convergence in distribution for a diffusively rescaled system of one - dimensional coalescing random walks starting at each point in the space and time lattice @xmath0 . to not get repetitive ,
when we make mention to the convergence of a system of random walks , we always consider one - dimensional random walks and diffusive space time scaling . in this introduction , we aim at describing informally the system and explain why its study is relevant . first , the limit is the brownian web ( bw ) which is a system of coalescing brownian motions starting at each point in the space and time plane @xmath1 .
it is the natural scaling limit of a system of simple coalescing random walks . here
, the bw is a proper random element of a metric space whose points are compact sets of paths .
this characterization of the bw was given by fontes , isopi , newman , ravishankar @xcite , although formal descriptions of systems of coalescing brownian motions had previously been considered , initially by arratia @xcite and then by tth , werner @xcite .
following @xcite , the study of convergence in distribution of systems of random walks to the bw and its variations has become an active field of research , for instance , see @xcite . in fontes ,
isopi , newman , ravishankar @xcite a general convergence criteria for the bw is obtained but only verified for a system of simple coalescing random walks that evolve independently up to the time of coalescence ( no crossing paths ) .
these criteria are used in newman , ravishankar , sun @xcite to prove convergence to the brownian web of coalescing nonsimple random walks that evolve independently up to the time of coalescence ( crossing paths ) .
coletti , fontes , dias @xcite have proved the convergenge of the bw for a system of coalescing random walks introduced by gangopadhyay , roy , sarkar @xcite .
this system is called the drainage network .
it evolves in the space time lattice @xmath0 in the following way : each site of @xmath0 is considered open or closed according to an iid family of bernoulli random variables ; if at time @xmath2 a walk is at site @xmath3 then at time @xmath4 its position is the open site nearest to @xmath3 ( if we have two possible choices for the nearest open site , we choose one of them with probability @xmath5 ) .
the drainage network is a system of coalescing nonsimple random walks with long range jumps where no crossing paths may occur .
futhermore the random walks paths are dependent before coalescence .
these properties make the convergence of the drainage network to the bw a relevant and nontrivial example .
we propose here a natural generalization of the drainage network and prove the convergence to the bw .
if at time @xmath2 a walk occupies the site @xmath3 , then it chooses at random @xmath6 and jumps to the @xmath7th open site nearest to @xmath3 , see the next section for the formal description .
now we have a system of coalescing nonsimple random walks with long range jumps and dependence before coalescence where crossing paths may occur .
we suggest to the reader the discussion on @xcite about the addicional difficulties that arise in the study of convergence of systems of coalescing random walks when crossing paths may occur .
we finish this section given a brief description of the bw which follows closely the description given in @xcite , see also @xcite and the appendix in @xcite . consider the extended plane @xmath8 ^ 2 $ ] as the completion of @xmath9 under the metric @xmath10 and let @xmath11 be the induced metric on @xmath12 . in @xmath13 , the lines @xmath14 \times \ { \infty \}$ ] and @xmath14 \times \ { - \infty \}$ ] correspond respectively to single points @xmath15 and @xmath16 , see picture 2 in @xcite .
denote by @xmath17 the set of all continuous paths in @xmath13 of the form @xmath18 \rightarrow ( f_\pi(t),t ) \in ( \bar{\mathbb{r}}^2,\bar{\rho})$ ] for some @xmath19 $ ] and @xmath20 \rightarrow [ -\infty,\infty ]
\cup \{\star \}$ ] . for @xmath21 , @xmath22 ,
define @xmath23 by @xmath24 we have a metric in @xmath17 such that @xmath25 is a complete separable metric space .
now define @xmath26 as the space of compact sets of @xmath25 with the topology induced by the hausdorff metric .
then @xmath26 is a complete separable metric space .
the brownian web is a random element @xmath27 of @xmath26 whose distribution is uniquely characterized by the following three properties ( see theorem 2.1 in @xcite ) : 1 . for any deterministic @xmath28 ,
almost surely there is a unique path @xmath29 of @xmath27 that starts at @xmath2 .
2 . for any finite deterministic set of points
@xmath30 , ... , @xmath31 in @xmath9 , the collection @xmath32 is distributed as coalescing brownian motions independent up to the time of coalescence .
3 . for any deterministic countable dense subset @xmath33 ,
almost surely , @xmath27 is the closure of @xmath34 in @xmath25 .
this paper is organized as follows : in section [ sec : model ] we describe the generalized drainage network model , obtain some basic properties and state the main result of this text which says that it converges to the bw .
the convergence is proved in section [ sec : convergence ] where fontes , isopi , newmann , ravishankar @xcite criteria is presented and discussed .
as we shall explain , a carefull analysis of the proof presented in @xcite reduce our work to prove an estimate on the distribution of the coalescing time between two walks in the model , which is carried in subsection [ subsec : coaltimes ] , and to prove that a diffusively rescaled system of @xmath35 walks in the model converge in distribution to @xmath35 coalescing brownian motions which are independent up to the coalescence time , this is done in subsection [ subsec : i ] .
let @xmath36 the set of positive integers . in order to present our model
we need to introduce some notation : * let @xmath37 be a family of independent bernoulli random variables with parameter @xmath38 and denote by @xmath39 the induced probability in @xmath40 .
* let @xmath41 be a family of independent bernoulli random variables with parameter @xmath5 and denote by @xmath42 the induced probability in @xmath40 .
* let @xmath43 be a family of independent and identically distributed random variables on @xmath36 with probability function @xmath44 $ ] and denote by @xmath45 the induced probability in @xmath46 .
we suppose that the three families above are independent of each other and thus they have joint distribution given by the product probability measure @xmath47 on the product space @xmath48 .
the expectation induced by @xmath49 will be denoted by @xmath50 .
some coupling arguments used in the proofs ahead will require enlargements of the probability space , we will keep the same notation since the formal specifications of these couplings are straightforward from their descriptions .
2 . at some points
, we will only be interested in the distributions of the sequences @xmath51 , @xmath52 and @xmath53 which do not depend on @xmath54 , justifying the following abuse of notation @xmath55 , @xmath56 and @xmath57 for @xmath54 . for points on the lattice
@xmath58 we say that @xmath59 is above @xmath60 if @xmath61 , immediately above if @xmath62 , at the right of @xmath60 if @xmath63 and at the left if @xmath64 .
moreover , we say that a site @xmath65 is open if @xmath66 and closed otherwise .
let @xmath67 be defined by putting @xmath68 as the @xmath7-th closest open site to @xmath2 which is immediately above it , when it is uniquely defined , otherwise @xmath69 is the @xmath7-th closest open site to @xmath2 which is immediately above and at the left of @xmath2 and @xmath70 is the @xmath7-th closest open site to @xmath2 which is immediately above and at the right of @xmath2 .
now for every @xmath65 define recursively the random path @xmath71 and @xmath72 for every @xmath73 .
let @xmath74 be the random directed graph with set of vertices @xmath75 and edges @xmath76 .
this generalizes the two - dimensional drainage network model introduced in @xcite which corresponds to the case @xmath77
. it will be called here the two - dimensional generalized drainage network model ( gdnm ) .
the random graph @xmath78 may be seen as a set of continuous random paths @xmath79 .
first , define @xmath80 , for every @xmath81 and @xmath82 .
the random path @xmath83 defined by linear interpolation as @xmath84 for every @xmath85 and @xmath86 . put @xmath87 where @xmath88 is the variance of @xmath89 ( at the end of the section we show that @xmath88 is finite if @xmath90 has finite second moment ) .
the random set of paths @xmath91 is called the generalized drainage network .
let @xmath92 for @xmath93 $ ] , be the diffusively rescaled generalized drainage network .
note that two paths in @xmath91 will coalesce when they meet each other at an integer time .
therefore , if @xmath94 , we have a system of coalescing random walks that can cross each other . our aim is to prove that @xmath95 converges in distribution to @xmath27 .
this extends the results in @xcite where the case @xmath77 was considered and no crossing between random walks occurs .
we say that @xmath90 has finite range if there exists a finite set @xmath96 such that @xmath97 if and only if @xmath98 .
our main result is the following : [ thm : conv ] if @xmath90 has finite range , then the diffusively rescaled gdnm , @xmath95 , converges in distribution to the @xmath27 as @xmath99 .
note that the hypothesis that @xmath90 has finite range does not imply that the random paths in the gdnm have finite range jumps .
it should be clear from the definition that any sufficiently large size of jump is allowed with positive probability .
we only need the hypothesis that @xmath90 has finite range at a specific point in the proof of theorem [ thm : conv](it will be discussed later ) , other arguments just require that @xmath90 has at most finite absolute fifth moment .
so we conjecture that theorem [ thm : conv ] holds under this condition .
the next section is devoted to the proof of theorem [ thm : conv ] .
we finish this section with some considerations about the random paths in the drainage network . the processes @xmath100 , @xmath65 , are irreducible aperiodic symmetric random walks identically distributed up to a translation of the starting point .
now consider @xmath101 as a random walk with the same distribution of @xmath102 .
we show that if @xmath90 has finite absolute m - th moment then @xmath103 also have this property .
[ prop : finmom ] if @xmath90 has finite absolute m - th moment then @xmath103 also has finite absolute m - th moment .
it is enough to show that the increments of @xmath104 have finite absolute m - th moment .
write @xmath105 for the increments of @xmath106 .
then @xmath107 is a sequence of iid random variables with symmetric probability function @xmath108 such that @xmath109 is equal to @xmath110 which is equal to @xmath111 q(k ) & { \nonumber}\end{aligned}\ ] ] therefore , since @xmath112 the absolute m - th moment of @xmath113 is bounded above by @xmath114 which is equal to @xmath115 the sum over @xmath2 is bounded by one half of the absolute m - th moment of a negative binomial random variable with parameters @xmath116 and @xmath117 which is a polynomial of degree @xmath118 on @xmath7
. therefore @xmath119 where @xmath120 is a polynomial of degree @xmath118 .
if @xmath90 has finite m - th moment then the right hand side above is finite , which means that @xmath108 also has finite m - th moment .
the proof of theorem [ thm : conv ] follows from the verification of four conditions given in @xcite for convergence to the brownian web of a system of coalescing random paths that can cross each other .
these conditions are described below : * let @xmath121 be a countable dense set in @xmath9 . for every @xmath122 , there exist single random paths @xmath123 , @xmath124 , such that for any deterministic @xmath125 , @xmath126 converge in distribution as @xmath127 to coalescing brownian motions , with diffusion constant @xmath128 depending on @xmath117 , starting at @xmath129 . *
let @xmath130 \textrm { and } z_2 \in [ -t , t ] \}$ ] and @xmath131 \textrm { and } z_2 \in [ t_0 , t_0+t]\}$ ] .
define @xmath132 as the event that there exists a path starting in the rectangle @xmath133 that exits the larger rectangle @xmath134 through one of its sides , where @xmath135 is a fixed constant .
then @xmath136 * let @xmath137 , @xmath138 , denote the random variable that counts the number of distinct points in @xmath139 that are touched by paths in @xmath140 which also touch some point in @xmath141\times \{t_0\}$ ] .
then , for every @xmath142 , @xmath143 * for every @xmath142 , @xmath144 for every @xmath142 . in @xcite a condition is derived to replace @xmath145 in the proof of convergence of diffusively rescaled non - nearest neighbor coalescing random walks to the brownian web .
their condition is based on duality and can be expressed as * let @xmath146 , @xmath138 , be the number distinct points in @xmath147 that are touched by a path that also touches @xmath148 .
then for any subsequential limit @xmath91 of @xmath95 we have that that @xmath149 \le \frac{b - a}{\sqrt{\pi t } } \ , .\ ] ] if we have proper estimate on the distributions of the coalescing times and condition i , then the proof of conditions @xmath150 , t and e follows from adaptations of arguments presented in @xcite , see also @xcite .
the estimate we need on the distribution of the coalescing time of two random walks in the drainage network starting at time zero is that the probability that they coalesce after time @xmath151 is of order @xmath152 .
this is central in arguments related to estimates on counting variables associated to the density of paths in the system of coalescing random walks under consideration .
let us be more precise . for each @xmath153 ,
let @xmath154 be the random path @xmath155 the processes @xmath156 are also spatially inhomogeneous random walks with mean zero square integrable increments , identically distributed up to a translation of @xmath2 and @xmath157 by the same amount .
for @xmath158 , let @xmath159 be random walk with the same distribution of @xmath160 .
define the random times @xmath161 and @xmath162 for every @xmath158 and @xmath163 .
[ density ] let @xmath169 be a system of coalescing random walks starting from every point of @xmath170 at time @xmath171 with increments distributed as the increments of @xmath106
. denote by @xmath172 the event that there is a walker seated at the origin by time @xmath151 . if ( [ eqprop : coaltime ] ) holds , then @xmath173 for some positive constant @xmath174 independent of everything else .
the proposition [ prop : coaltime1 ] , lemma [ prop : coaltime2 ] and lemma [ density ] are respectively versions of lemmas 2.2 , 2.4 and 2.7 in section 2 of @xcite .
their results hold for the difference of two independent random walks what is not our case .
the proposition [ prop : coaltime1 ] is by itself a remarkable result in our case and is going to be proved in section [ subsec : coaltimes ] . in the proof , we need @xmath90 with finite range due to an estimate to bound the probability that two walks in the gdnm coalesce given they cross each other ( item ( iii ) of lemma [ lemma : coupling ] ) .
it is the only place where the finite range hypothesis is used .
both lemmas follow from inequality ( [ eqprop : coaltime ] ) in proposition [ prop : coaltime1 ] following arguments presented in @xcite , although we also prove lemma [ prop : coaltime2 ] in section [ subsec : coaltimes ] to make clear that it holds in our case . concerning the proof of lemma [ density ] , it is exactly the same as the proof of lemma 2.7 in @xcite . now let us return to the analysis of conditions @xmath150 ,
t and e. the idea we carry on here is that if convergence to the bw holds for a system with non - crossing paths then the proof of @xcite can be adapted to similar systems allowing crossing paths if we have proposition [ prop : coaltime1 ] and condition @xmath175 . in this direction
, @xcite is extremly useful since it is carefully written in a way that for each result one can clearly identify which hypotheses are needed to generalize its proof to other systems of coalescing random walks .
the proof of @xmath150 is the same as that of section 3 in @xcite for systems of coalescing random walks independent before coalescence . in order to reproduce the proof given in @xcite
we need finite second moment of @xmath103 and that the coalescing time of two walkers starting at distance @xmath176 apart is of order @xmath152 which is the content of proposition [ prop : coaltime1 ] .
the proof of condition t is also the same as that of section 4 in @xcite for systems of coalescing random walks independent before coalescence .
the proof depends only on two facts : first , an upper bound for the probability of the time it takes for two random walks to get far away one from each other before they coalesce , which is the content of lemma [ prop : coaltime2 ] ; second , finite fifth moment of the increments of the random walks , which follows from the finite fifth moment of @xmath90 .
now we consider condition @xmath177 , but first we introduce more notation . for an @xmath178valued random variable @xmath179 ,
define @xmath180 to be the subset of paths in @xmath179 which start before or at time @xmath151 . also denote by @xmath181 the set of values at time @xmath151 of all paths in @xmath179 .
in @xcite it is shown that condition @xmath177 is implied by condition @xmath182 stated below : * if @xmath183 is any subsequential limit of @xmath184 for any @xmath185 for any non - negative sequence @xmath186 going to zero as @xmath35 goes to infinity , then @xmath187 with @xmath188 and @xmath189 \le e[\hat{\eta}_{\mathcal{w}}(t_0,t;a , b ) ] = \frac{b - a}{\sqrt{\pi t}}.$ ] the verification of condition @xmath190 follows using the same arguments presented in section 6 of @xcite . denote by @xmath191 the subset of paths in @xmath183 which start before or at time @xmath192 and truncated before time @xmath193 . by lemma [ density ]
we get the locally finiteness of @xmath194 for any @xmath195 ( for more details see @xcite , @xcite ) . the key point in the verification of condition
@xmath190 is to note that for any @xmath196 \le e[\hat{\eta}_{\mathcal{z}_{t_0}^{(t_0 + \epsilon)_t}}(t_0 + \epsilon , t - \epsilon;a , b)]$ ] .
then , by the locally finiteness of @xmath197 and condition @xmath175 we get that @xmath191 is distributed as coalescing brownian motions starting from the random set @xmath194 .
we get condition @xmath190 from the fact that a family of coalescing brownian motions starting from a random locally finite subset of the real line is stochastically dominated by the family of paths of coalescing brownian motions starting at every point in @xmath198 at time @xmath199 and taking limit when @xmath200 goes to zero .
see lemma 6.3 in @xcite that also holds in our case .
we start this section proving lemma [ prop : coaltime2 ] assuming that proposition [ prop : coaltime1 ] holds .
we are going to follow closely the proof of proposition 2.4 in @xcite aiming at making clear that their arguments holds in our case .
their first estimate is @xmath201 holds for @xmath202 sufficiently small .
it follows from the following estimate based on the strong markov property , which in our case also holds by the independence of the increments , @xmath203 where @xmath204 is the set of trajectories that remain in the interval @xmath205 $ ] during the time interval @xmath206 $ ] . for every @xmath163 and @xmath166 , write @xmath207
then @xmath208 where @xmath209 is a standard normal random variable , the last inequality is a consequence of donsker invariance principle , see lemma 2.3 in @xcite .
therefore we have that lemma [ prop : coaltime2 ] .
the remain of the section is entirely devoted to the proof of proposition [ prop : coaltime1 ] .
it suffices to consider the case in which @xmath210 , see the proof of lemma 2.2 in @xcite . here
we will considering that @xmath90 has finite second moment except near the end of the section , in the proof of condition ( iii ) in lemma [ lemma : coupling ] , where we need @xmath90 with finite range .
the proof is based on a proper representation of the process @xmath211 which has already been used in @xcite . however , in our case , the possibility to have @xmath212 assuming negative values , due to crossing paths , requires a new approach . to simplify notation , we write @xmath213 . by skorohod representation theorem ,
see section 7.6 in @xcite , there exist a standard brownian motion @xmath214 and stopping times @xmath215 , @xmath216 , ... , such that @xmath217 has the same distribution of @xmath218 , for @xmath219 , where @xmath220 .
furthermore , the stopping times @xmath215 , @xmath216 , ... , have the following representation : @xmath221 where @xmath222 is a family of independent random vectors taking values in @xmath223 .
also , by the definition of @xmath106 , we have that for each fixed @xmath224 , @xmath225 , @xmath73 , are identically distributed .
indeed , write @xmath226 for the increments of @xmath227 , then @xmath228 and the families of identically distributed ( but dependent ) random variables @xmath229 , @xmath54 , are independent of each other .
moreover @xmath230 therefore , by proposition [ prop : finmom ] , @xmath231 has finite second moment uniformly bounded in @xmath224 . also note that , by symmetry , for @xmath232 , @xmath231 has the same distribution as @xmath233 .
let us give a brief description of the idea behind the proof of proposition [ prop : coaltime1 ] .
by similar arguments as those of @xcite we show that @xmath234 is of order @xmath152 , for some @xmath235 sufficiently small . on the other hand
if @xmath236 , then for @xmath151 large , we show that @xmath237 is enough time for the brownian excursions to guarantee that @xmath238 have sufficient sign changes such that , under these changes , the probability that @xmath238 does not hit @xmath171 is of order @xmath152 .
now define the random discrete times @xmath239 , @xmath240 and for @xmath241 @xmath242 then @xmath243 is a increasing sequence of stopping times for the random walk @xmath211 , which is strictly increasing until @xmath244 such that @xmath245 .
note that @xmath246 if @xmath244 is even and @xmath247 if @xmath244 is odd .
the sequence @xmath243 induce the increasing sequence of continuous random times @xmath248 .
we have the inequality @xmath249 where @xmath250 .
we are going to show that both probabilities in the rightmost side of ( [ ineqct ] ) are of order @xmath152 . during the proof we will need some technical lemmas whose proofs
are postponed to the end of this section .
we start with @xmath251 .
note that we can write @xmath252 where @xmath253 are independent random variables .
the sequences @xmath254 , indexed by @xmath255 , are not identically distributed .
fix @xmath256 . by the markov inequality we have @xmath257 as in @xcite ,
put @xmath258 , then we have that @xmath259^t .\ ] ] to estimate the expectation above , we need some uniform estimates on the distribution of @xmath260 and @xmath261 .
let @xmath265 be the exit time of interval @xmath266 by a standard brownian motion .
we have that @xmath267 where @xmath268 is given by lemma [ p00 ] and @xmath269 .
now , choose @xmath270 such that @xmath271 < 1 $ ] , then from ( [ eq : zeta1 ] ) and ( [ eq : zeta2 ] ) we have @xmath272 where @xmath273 .
finally , choose @xmath274 such that @xmath275 and @xmath276 for @xmath166 . 1 .
@xmath281 are iid random variables ; 2 .
@xmath282 has the same distribution as @xmath283 ; 3 .
@xmath284 ; 4 .
@xmath285 is stochastically dominated by @xmath286 which is defined as @xmath287 , @xmath288 and @xmath289 where @xmath214 is a standard brownian motion independent from the sequence @xmath290 ; 5 .
@xmath291 implies @xmath292 which is equivalent to @xmath293 given that @xmath294 .
let @xmath295 be as in the statement of lemma [ lemma : coupling ] .
since @xmath296 is a subset of @xmath297 by lemma [ lemma : coupling ] , we have that ( [ eq : tjestimate1 ] ) is bounded above by @xmath298 for the right hand side above , write @xmath299 where the last equality follows from the independence of the @xmath279 s . now put @xmath300 and define @xmath301 and @xmath302 then @xmath303 is also a sequence of iid square integrable random variables and , from lemma [ lemma : coupling ] , we get that @xmath304 to estimate the right hand side in the previous equality , write @xmath305 then it is easy to verify that @xmath306 , @xmath280 , are identically distributed random variables which are not independent .
however @xmath307 and @xmath308 are families of iid random variables . fix @xmath322 .
then @xmath323 since the last term in the previous inequality does not depend on @xmath118 , we have that @xmath324 to prove the other inequality we consider two cases .
first , suppose that @xmath325 , then @xmath326 now we consider the case @xmath327 .
in this case @xmath328 therefore , @xmath329 _ proof of lemma [ lemma : coupling ] .
_ consider independent standard brownian motions @xmath214 and @xmath331 such that @xmath214 is used in the skorohod representation of @xmath211 and @xmath331 is also independent of @xmath211 . in the time interval
@xmath332 $ ] , @xmath214 makes an excursion being able to visit @xmath333 $ ] when @xmath334 for some @xmath335 .
when this happens two things may occur : 1 .
@xmath336 which implies @xmath337 and @xmath338 meaning that @xmath339 and @xmath340 have coalesced ; 2 .
@xmath341 which implies that , with probability greater than some strictly positive constant @xmath342 , @xmath337 and @xmath214 will leave the interval @xmath343 $ ] by its left side ( strong markov property with the fact that @xmath214 visits @xmath344 ) meaning that @xmath339 and @xmath345 have crossed .
denote @xmath346 by the random variable that denotes the cardinality of @xmath347 \textrm { in time interval } [ t_{n-1 } , t_{n } ] \big\}$ ] .
note that @xmath348 and , from ( 2 ) , @xmath346 is stochastically bounded by a geometric random variable with parameter @xmath342 .
we will construct below a sequence of iid non positive square integrable random variables @xmath349 such that for each @xmath350 , @xmath351 .
define @xmath352 where @xmath353 is a geometric random variable with parameter @xmath342 such that @xmath354 .
it is a simple to show that @xmath355 is square integrable and independent of @xmath331 .
clearly from the definitions @xmath356 so if we take @xmath283 with the same distribution of @xmath357 then the time for @xmath214 starting at @xmath176 to hit @xmath358 is stochastically dominated by the time for @xmath331 starting at @xmath283 to hit @xmath358 which we denote @xmath359 . from this point
, it is straightforward to use an induction argument to build the sequence @xmath360 . at step @xmath244 in the induction argument
, we consider the @xmath214 excursion in time interval @xmath361 $ ] , and since @xmath362 we can obtain @xmath279 and define @xmath286 using @xmath331 as before . by the strong markov property of @xmath363
, we obtain that the @xmath279 s are independent and properties ( i ) , ( ii ) , ( iv ) and ( v ) follows directly from the construction .
* construction of @xmath349 : * fix @xmath73 .
we have to show that there exists a square integrable random variable @xmath364 that stochastically dominates @xmath365 for every @xmath366 . from this
it is straightforward to obtain the sequence @xmath349 , by induction and extension of the probability space , in a way that @xmath367 has the same distribution as @xmath364 .
now fix @xmath366
. we will be conditioning of the event @xmath368 . recall that @xmath369 , @xmath86 .
if @xmath370 and @xmath371 were independent , @xmath211 would be spatially homogeneous and @xmath364 could be fixed as the sum of the absolute values of two independent increments of the processes @xmath372 , @xmath373 .
that is not the case .
we define @xmath374 if @xmath375 . on the event @xmath376
, @xmath364 should be smallest value that the environment at time @xmath377 allows for @xmath378 .
so let us regard the environment at time @xmath377 .
consider the partition of the sample space on the events @xmath379 , for @xmath232 and @xmath158 .
we define random variables @xmath380 , @xmath381 and @xmath382 in the following way : on @xmath383 @xmath384 @xmath385 @xmath386 see figure @xmath176 below . clearly , conditioned to each @xmath383 , @xmath380 , @xmath381 and @xmath382 are iid whose distribution is caracterized as the sum of @xmath209 geometric random variables , where @xmath209 is distributed according to the probability function @xmath90 and is independent of the geometric random variables .
in particular , @xmath380 , @xmath381 and @xmath382 do not depend on @xmath7 and @xmath118 and it is straightforward to verify that they have finite second moment if @xmath90 does .
define @xmath387 . on the event @xmath376
, @xmath388 can not be smaller than @xmath364 .
indeed on @xmath383 , if the random walks @xmath339 and @xmath345 cross each other , we have @xmath389 thus @xmath390 in figure @xmath176 above the interval with size @xmath380 has exactly @xmath393 open sites , the interval with size @xmath381 has @xmath393 open sites and finally the interval with size @xmath382 has exactly @xmath394 open sites . *
proof of ( iii ) : * conditioned on the event @xmath395 , which occurs with probability @xmath342 , we have that @xmath396 and we can suppose @xmath355 equal in distribution to @xmath364 defined above .
thus @xmath397 where the first inequality follows from the definition of @xmath364 .
so we have ( ii ) if @xmath398 note that the probability above , is the probability that two walks in the gdnm at distance @xmath118 coalesce after their first jumps given that they cross each other .
lets prove ( [ eq : cruzaecoal ] ) .
suppose that @xmath90 has finite range ( here is the single point in the text where we need @xmath90 with finite range ) .
fix @xmath399 such that @xmath400 , if @xmath401 .
since clearly @xmath402 for every @xmath366 , it is enough to show that @xmath403 then fix @xmath404 and write @xmath405 where by symmetry the left hand side is equal to @xmath406 we claim that each term in the previous sum is bounded above by @xmath407 , for some @xmath164 not depending on @xmath118 , which implies that @xmath408 let us consider @xmath409 first ..
if @xmath410 and @xmath411 , we define @xmath412 clearly @xmath413 and @xmath414 and we can choose @xmath415 such that @xmath416 for @xmath417 and if @xmath418 than @xmath419 . by changing the occupancies at sites @xmath420 we have that the set @xmath421 has probability bounded above by @xmath422 therefore we have proved the claim for the first probability in ( [ eq : dec1 ] ) .
now we estimate @xmath423 .
since @xmath424 , if @xmath410 , @xmath425 and @xmath426 , then there exists at least @xmath427 open sites in the interval @xmath428 .
by changing the occupancies of at most @xmath427 of these sites , we get a configuration with @xmath429 .
thus , @xmath423 is bounded above by @xmath430 [ prop : coalbm ] suppose that inequality ( [ eqprop : coaltime ] ) holds and @xmath90 has finite absolute second moment .
let @xmath434 , @xmath435 , ...
, @xmath436 be @xmath437 distinct points in @xmath9 such that @xmath438 , and if @xmath439 for some @xmath440 , @xmath441 , then @xmath442 .
then @xmath443 where @xmath444 are coalescing brownian motions with constant diffusion coefficient @xmath128 starting at @xmath445 .
let @xmath447 be a vector of coalescing brownian motions starting at @xmath448 with diffusion coefficient @xmath128 . under the assumption that @xmath449 is finite range , we show in this section that @xmath450 we have that the convergence of @xmath451 to a brownian motion starting at @xmath3 is a direct consequence of donsker invariance principle .
for the case @xmath77 , in the proof presented in @xcite , it is shown by induction that for @xmath452 @xmath453 it relies on the fact that no crossing can occur which allows the use of a natural order among trajectories of the random walks which we do not have .
we will take a slight different approach here .
we make the proof in the case @xmath454 , @xmath455 , ... ,
@xmath456 , the other cases can be carried out in an analogous way . to simplify the notation write @xmath457
let us fix a uniformly continuous bounded function @xmath458)^{m+1 } { \rightarrow}\mathbb{r}$ ] .
we show that @xmath459 - \textrm{e}[h(b^{0}_0 , \ , \ldots \ , , b^{m}_m)]| = 0 \ , .\end{aligned}\ ] ] as mentioned above @xmath460 converges in distribution to @xmath461 .
now , we suppose that @xmath462 converges in distribution to @xmath463 and we are going to show that ( [ i2 ] ) holds . by induction and the definition of convergence in distribution
we obtain ( [ eq : convdist1 ] ) .
we start defining a modifications of the random walks @xmath464 , ... , @xmath465 ,
@xmath466 in order that the modification of @xmath466 is independent of the modifications of @xmath464 , ... , @xmath465 , until the time of coalescence with one of them .
we achieve this through a coupling which is constructed using a suitable change of the environment .
so let @xmath467 and @xmath468 be two independent families of independent bernoulli random variables with parameter @xmath38 which are also independent of @xmath469 , @xmath470 and @xmath471 . considering the processes @xmath472 , ... ,
@xmath473 we define the environment @xmath474 by @xmath475 the processes @xmath476 , ... , @xmath477 are defined in the following way : for every @xmath478 , @xmath479 starts at @xmath480 and evolves in the same way as @xmath481 except that @xmath482 sees a different environment .
both @xmath481 and @xmath482 processes use the families of random variables @xmath470 and @xmath471 , but @xmath482 jumps according to @xmath483 until the random time @xmath151 such that @xmath484 after that time @xmath482 jumps according to @xmath467 . from @xmath482
we define @xmath485 and @xmath486 as before . considering the process @xmath487 we define the environment @xmath488 by @xmath489
the process @xmath490 is defined in the following way : @xmath490 starts at @xmath491 and evolves in the same way as @xmath487 exept that @xmath490 sees a different environment . both @xmath487 and @xmath490
use the families of random variables @xmath470 and @xmath471 , but @xmath490 jumps according to @xmath483 until the random time @xmath151 such that @xmath492 , \ ] ] after that time @xmath490 jumps according to @xmath468 . from @xmath490
we define @xmath493 and @xmath494 as before .
* @xmath490 is independent of @xmath496 when conditioned to the event that @xmath497 do not get to a distance smaller than @xmath498 of some @xmath499 , @xmath500 .
* for every @xmath478 , @xmath481 and @xmath482 are equal at least until the first jump by one of them of size greater than @xmath501 , thus they are equal when retricted to the event @xmath502 .
the same is true about @xmath487 and @xmath490 . *
@xmath503 and @xmath462 have the same distribution .
we still need to replace the process @xmath490 by a process @xmath507 in such a way that @xmath490 is independent of @xmath508 until the time @xmath507 coalesce with one of them . in order to do this
we define @xmath507 in the following way : let @xmath509 then @xmath510 , for @xmath511 , otherwise @xmath512 jumps according to @xmath468 or coalesce with one of @xmath513 , ... , @xmath477 , if they meet each other at an integer time . from @xmath512
we define @xmath514 and @xmath515 as before .
define the event @xmath516 and note two facts : we are now ready to prove ( [ i2 ] ) . write @xmath520 - \textrm{e}[h(b^{0}_0 , \ , \ldots \
, , b^{m}_m)]| \le } { \nonumber}\\ & & \qquad |\textrm{e}[h(\delta^{0}_n , \ , \ldots \ , , \delta^m_n ) ] - \textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \
, , \tilde{\delta}^{m-1}_n , \hat{\delta}^m_n)]| { \nonumber}\\ & & \qquad \quad + |\textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \hat{\delta}^m_n ) ] - \textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \bar{\delta}^m_n)]| \\ & & \qquad \quad \quad + |\textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \bar{\delta}^m_n ) ] - \textrm{e}[h(b^{0}_0 , \ , \ldots \ , , b^{m}_m)]| . { \nonumber}\end{aligned}\ ] ] by donsker s invariance principle and the induction hypothesis
, we have that @xmath521 - \textrm{e}[h(b^{0}_0 , \ , \ldots \
, , b^{m}_m)]| = 0 \ , .\ ] ] so we only have to deal with the first and second term in ( [ i1 ] ) .
for the first term in ( [ i1 ] ) we have that @xmath522 - \textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \hat{\delta}^m_n)]| \le } \\ & & \qquad = |\textrm{e } [ ( h(\delta^{0}_n , \ , \ldots \ , , \delta^m_n ) - h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \hat{\delta}^m_n ) ) \mathbb{i}_{\mathcal{b}^c_{n , s}}]| \\ & & \qquad \le c \|h\|_{\infty}\ , n^{-\frac{1}{4 } } \ , .
\end{aligned}\ ] ] hence @xmath523 - \textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \hat{\delta}^m_n)]| = 0 \ , .\ ] ] it remains to prove that the second term in ( [ i1 ] ) converges to zero as @xmath35 goes to @xmath524 .
note that @xmath525 - \textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \bar{\delta}^m_n)]| = } \\ & & \qquad = |\textrm{e}[h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \hat{\delta}^m_n ) - h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \bar{\delta}^m_n)]| \\ & & \qquad = |\textrm{e } [ ( h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n ,
\hat{\delta}^m_n ) - h(\tilde{\delta}^{0}_n , \ , \ldots \ , , \tilde{\delta}^{m-1}_n , \bar{\delta}^m_n ) ) \mathbb{i}_{\mathcal{a}^c_{n , s } } ] | \end{aligned}\ ] ] the rightmost expression in the previous equality is bounded above by @xmath526 plus @xmath527|\ ] ] which is equal to @xmath528 for every @xmath529 .
the times @xmath530 and @xmath531 have the tail of their distributions of @xmath532 . also define @xmath533 note that on @xmath534 we have @xmath535 , for some @xmath536 .
furthermore , on @xmath502 and up to time @xmath537 , we have @xmath538 equal to @xmath539 .
we have that @xmath540 is equal to @xmath541 for each @xmath542 , the respective term in the previous sum is bounded above by @xmath543 \ } \big ) { \nonumber}\\ & & \textrm{p } \big ( \ { \tau_k > \nu_k + n^{\frac{7}{4 } } \ } \cup \ { \bar{\tau}_k > \nu_k + n^{\frac{7}{4 } } \ } \big ) .\end{aligned}\ ] ] the second term in ( [ i3 ] ) is bounded above by @xmath544 and the first by @xmath545 which by donsker invariance principle goes to zero as @xmath35 goes to infinity .
finally we have that ( [ i4 ] ) is bounded above by a term that converges to zero as @xmath546 plus @xmath547 \big| \ , , \ ] ] where @xmath548 by the uniform continuity of @xmath549 the rightmost expectation in the previous expression converges to zero as @xmath35 goes to @xmath550 . | we introduce a system of one - dimensional coalescing nonsimple random walks with long range jumps allowing crossing paths and exibiting dependence before coalescence .
we show that under diffusive scaling this system converges in distribution to the brownian web . |
consider the following identity : @xmath93\nonumber\\ & = & g(\tilde \delta ) \left[1+\frac{\alpha}{g(\tilde \delta ) } ( y^+ - \tilde \delta)\right ] \ , \end{aligned}\ ] ] where @xmath94 next introduce newly renormalized units using the effective viscosity @xmath95 , i.e. @xmath96 in terms of these variables the balance equations are rewritten as @xmath97s^\ddag+ w^\ddag = 1\ , \label{vepr1}\\ & & [ 1+\alpha(y^\ddag-\delta^\ddag)]\frac{{\delta}^2(\alpha)}{{y^\ddag}^2 } + \frac{\sqrt{w^\ddag}}{\kappa_{_{\rm k } } y^\ddag } = s^\ddag \ .
\label{vepr2}\end{aligned}\ ] ] these equations are isomorphic to ( [ pr1 ] ) and ( [ pr2 ] ) with @xmath62 replaced by @xmath98 . the ansatz ( [ scaling ] ) is then replaced by @xmath99 this form is dictated by the following considerations : ( i ) @xmath65 when @xmath66 , ( ii ) all lengths scales in the rescaled units are divided by @xmath95 , and thus the pre - factor in front of @xmath100 becomes @xmath101 , and ( iii ) @xmath102 in eq .
( [ bal2 ] ) is now replaced in eq .
( [ vepr2 ] ) by @xmath103 , leading to the new argument of @xmath100 . since the function @xmath70 can not change due to the change of variables , the function @xmath70 given by eq .
( [ delta2 ] ) should be identical to that given by eq .
( [ scaling ] ) : @xmath104 using the explicit form of @xmath95 eq .
( [ gt ] ) , and choosing ( formally first ) @xmath105 we find that @xmath106 .
it is easy to verify that this is indeed the solution of the above equation for any value of @xmath98 , and therefore the unique form of eq .
( [ delta ] ) .
we thank t.s . lo and anna pomyalov for very useful discussions , and for careful reading of early versions of the manuscript .
this work has been supported in part by the us - israel binational science foundation , by the european comission under a tmr grant , and by the minerva foundation , munich , germany . | drag reduction by polymers in wall turbulence is bounded from above by a universal maximal drag reduction ( mdr ) velocity profile that is a log - law , estimated experimentally by virk as @xmath0 . here
@xmath1 and @xmath2 are the mean streamwise velocity and the distance from the wall in wall " units . in this letter
we propose that this mdr profile is an edge solution of the navier - stokes equations ( with an effective viscosity profile ) beyond which no turbulent solutions exist .
this insight rationalizes the universality of the mdr and provides a maximum principle which allows an ab - initio calculation of the parameters in this law without any viscoelastic experimental input .
the mean streamwise velocity profile in newtonian turbulent flows in channel geometries satisfies the classic von - krmn log - law of the wall " which is written in wall units as @xmath3 here @xmath4 , @xmath5 and @xmath6 are the streamwise , wall - normal and spanwise directions respectively @xcite .
the wall units are defined as follows : let @xmath7 be the fixed pressure gradients @xmath8 , and @xmath9 the mid - height of the channel .
then the reynolds number , the normalized distance from the wall @xmath2 and the normalized mean velocity @xmath10 ( which is in the @xmath4 direction with a dependence on @xmath5 only ) are defined by @xmath11 where @xmath12 is the kinematic viscosity .
the law ( [ karman ] ) is universal , independent of the nature of the newtonian fluid ; it is one of the shortcomings of the theory of wall - bounded turbulence that the von - krmn constant @xmath13 and the intercept @xmath14 are only known from experiments and simulations @xcite .
one of the most significant experimental findings @xcite concerning turbulent drag reduction by polymers is that in channel and pipe geometries the velocity profile ( with polymers added to the newtonian fluid ) is bounded between von - krmn s log - law and another log - law which describes the maximal possible velocity profile ( maximum drag reduction , mdr ) @xcite , @xmath15 this law , which had been discovered experimentally by virk ( and hence the notation @xmath16 ) , is also claimed to be universal , independent of the newtonian fluid and the nature of the polymer additive , including flexible and rigid polymers @xcite .
the numerical value of the coefficient @xmath16 is presently known only from experiments , @xmath17 , giving a phenomenological mdr law in the form @xcite @xmath18 for sufficiently high values of and concentration of the polymer , the velocity profile in a channel is expected to follow the law ( [ final ] ) . for finite , finite concentration and finite extension of the polymers one expects cross - overs back to a velocity profile parallel to the law ( [ karman ] ) , but with a larger mean velocity ( i.e. with a larger value of the intercept @xmath19 ) .
the position of the cross - overs are not universal in the sense that they depend on the nature of the polymers and the flow conditions ; the cross - overs are discussed in @xcite .
while we still can not predict from first principles the parameters in von - krmn s log - law , the aim of this letter is to identify the mdr log - law as an edge turbulent state in wall bounded flows , leading to a derivation of the parameters appearing in eq .
( [ finalexp ] ) without any viscoelastic input .
the derivation follows from the theory of drag reduction by polymers that was developed recently , and therefore a short summary of the main aspects of the theory is in order .
wall bounded turbulence in newtonian fluids is discussed @xcite by considering the fluid velocity @xmath20 as a sum of its average ( over time ) and a fluctuating part : @xmath21 the objects that enter the newtonian theory are the mean shear @xmath22 , the reynolds stress @xmath23 and the kinetic energy @xmath24 : @xmath25 in the presence of dilute polymers added to the newtonian fluid one needs to complement these statistical objects with the dimensionless
conformation tensor " @xmath26 which stems from the ensemble average of the dyadic product of the end - to - end distance of the polymer chains , normalized by its equilibrium value @xmath27 @xcite .
the way that the conformation tensor appears in the additional stress tensor which appears in the viscoelastic equations of motion is model dependent ; it is different for flexible and rigid polymers , and it also depends on the actual model of the polymers .
nevertheless it was shown theoretically @xcite that both for rigid and flexible polymers one can write down eventually the balance equations for momentum and energy at distance @xmath5 away from the wall as @xmath28k(y)&= & w(y)s(y ) \ , \label{finalbalevis}\\ \nu(y ) & \equiv & \nu_0 + c \nu_p \langle r_{yy}\rangle \ .
\label{eeffvis}\end{aligned}\ ] ] in eq.([finalmom ] ) the right hand side is the rate at which momentum is generated by the pressure head , @xmath23 is the momentum flux in physical space towards the wall , and @xmath29 stands for the newtonian viscous dissipation of momentum in addition to the polymer contribution to the dissipation of momentum . the effective viscosity @xmath30 is given by eq . ( [ eeffvis ] ) where @xmath31 is the viscosity due to the polymers in the limit of zero shear and @xmath32 is a constant of the order of unity .
similarly , in eq .
( [ finalbalevis ] ) the first term on the left hand side is the combined newtonian and polymer contributions to the energy dissipation , the second models the inertial energy fluxes , and the right hand side is the ( exact ) energy production rate .
the coefficients @xmath33 and @xmath34 are dimensionless and of the order of unity . as in the newtonian case ,
the balance equations need to be supplemented by a relation between @xmath24 and @xmath23 .
rigorously the cauchy - schwartz ineqality leads to @xmath35 ; experimentally one find that @xmath36 with @xmath37 apparently @xmath5-independent outside the viscous boundary layer . to derive the functional form of the mdr
@xcite one asserts that the terms containing @xmath30 in the balance equations ( [ finalmom ] ) and ( [ finalbalevis ] ) overwhelm the inertial terms , and then together with ( [ wk ] ) one derives @xmath38 which is the log - law for the mdr .
consisten with this law the effective viscosiy turns to be linear in @xmath5 .
the increase in the viscosity of course increases the dissipation , but it was argued that the momentum flux @xmath39 is reduced even further , leading to an increase in the mean momentum of the flow ( for a given pressure head ) , i.e. drag reduction .
it is easy to argue @xcite that the slope of the new log - law is larger that the slope in von - krmn s log - law , and hence drag reduction is obtained .
nevertheless , since neither @xmath37 nor the constants @xmath33 and @xmath34 in ( [ finalbalevis ] ) are known apriori , the actual slope of the mdr could not be determined .
this shortcoming is remedied in the rest of this letter . the crucial new insight that will explain the universality of the mdr and furnish the basis for its calculation
is that the mdr is a marginal flow state of wall - bounded turbulence . in other words
this is the solution of eqs .
( [ finalmom ] ) and ( [ finalbalevis ] ) for which @xmath22 ( or equivalently , the velocity profile ) is the maximal possible for any turbulent solution . attempting to increase @xmath40 further results in the collapse of the turbulent solutions in favor of a stable laminar solution @xmath41 . as such
, the mdr is universal by definition , and the only question is whether a polymeric ( or other additive ) can supply the particular effective viscosity @xmath30 that drives eqs .
( [ finalmom ] ) and ( [ finalbalevis ] ) to attain the marginal solution that maximizes the velocity profile .
we predict that the same marginal state will exist in numerical solutions of the navier - stokes equations furnished with a @xmath5-dependent viscosity @xmath30 .
there will be no turbulent solutions with velocity profiles higher than the mdr . to see this explicitly , we first rewrite the balance equations in wall units . define @xmath42 , taken for simplicity as @xmath5-independent
; we know from the newtonian limit ( in which @xmath43 ) that @xmath44 @xcite .
once we write the equations with @xmath30 the ratio @xmath45 can change drastically , and we denote it below by @xmath46 . with @xmath47 , @xmath48 , @xmath49 and @xmath50 ,
the balance equations are written as @xmath51 in eq .
( [ bal2 ] ) @xmath52 when @xmath53 ( the newtonian limit ) . the bunch of numerical constants in the second term on the lhs of ( [ bal2 ] )
was replaced with @xmath54 in agreement with newtonian log - law when @xmath53 .
in fact the second term on the left hand side of eq.([bal2 ] ) ( which vanishes at the mdr ) contains a factor @xmath55 .
this factor is omitted for simplicity ; accounting for this factor complicates slightly the algbra , leaving the final conclusions unchanged .
substituting now @xmath56 from eq .
( [ bal1 ] ) into eq .
( [ bal2 ] ) leads to a quadratic equation for @xmath57 .
this equation has as a zero solution for @xmath58 ( laminar solution ) as long as @xmath59 .
turbulent solutions are possible only when @xmath60 .
thus at the edge of existence of turbulent solutions we find @xmath61 .
this is not surprising , since it was observed already in previous work that the mdr solution is consistent with an effective viscosity which is asymptotically linear in @xmath2 @xcite .
it is therefore sufficient to seek the edge solution by maximizing the velocity profile with respect to linear viscosity profiles , and we rewrite eqs .
( [ bal1 ] ) and ( [ bal2 ] ) with an effective viscosity that depends linearly on @xmath2 outside the boundary layer of thickness @xmath62 : @xmath63s^+ + w^+ = 1\ , \label{pr1}\\ & & [ 1+\alpha(y^+-\delta^+)]\frac{{\delta}^2(\alpha)}{{y^+}^2 } + \frac{\sqrt{w^+}}{\kappa_{_{\rm k } } y^+ } = s^+\ .
\label{pr2}\end{aligned}\ ] ] we now endow @xmath64 with an explicit dependence on the slope of the effective viscosity , @xmath65 when @xmath66 .
for @xmath67 the ratio @xmath68 is expected to depend on @xmath69 ( drag reduction involves a reduction in @xmath39 ) , and this dependence is an important part of the following theory .
we can present @xmath70 in terms of a dimensionless scaling function @xmath71 , @xmath72 obviously , @xmath73 . in the appendix
we show that the balance equation ( [ pr1 ] ) and ( [ pr2 ] ) ( with the prescribed form of the effective viscosity profile ) have an non - trivial symmetry that leaves them invariant under rescaling of the wall units .
this symmetry dictates the function @xmath70 in the form @xmath74 armed with this knowledge we can now find the maximal possible velocity far away from the wall , @xmath75 .
there the balance equations simplify to @xmath76 these equations have the @xmath2-independent solution for @xmath57 and @xmath77 : @xmath78 obviously , ( see fig .
[ fig1 ] ) , the supremum of @xmath77 is obtained when @xmath58 vanishes , which happens precisely when @xmath79 .
using eq .
( [ delta ] ) we find the solution @xmath80 .
then @xmath81 , giving @xmath82 . using the estimate @xmath44 we get the final prediction for the mdr . using eq .
( [ final ] ) with @xmath83 , we get @xmath84 this result is in close agreeement with the empirical law ( [ finalexp ] ) proposed by virk . note that the numbers appearing in virk s law correspond to @xmath85 , which is well within the error bar on the value of this newtonian parameter . note
that we can easily predict where the asymptotic law turns into the viscous layer upon the approach to the wall .
we can consider an infinitesimal @xmath58 and solve eqs .
( [ bal1 ] ) and ( [ bal2 ] ) for @xmath56 and the viscosity profile .
the result , as before , is @xmath86 . since the effective viscosity can not fall bellow the newtonian limit @xmath87 we see that the mdr can not go below @xmath88 .
we thus expect an extension of the viscous layer by a factor of 2 .
one should note that the result @xmath89 should not be interpreted as @xmath41 .
the differeence between the two objects is the factor of @xmath90 , @xmath91 .
since the mdr is reached asymptotically as @xmath92 , there is enough turbulence at this state to stretch the polymers to supply the needed effective viscosity .
nevertheless our discussion is in close correspondence with the experimental remark by virk @xcite that close to the mdr asymptote the flow appears laminar . in summary
, one can probably improve further the model of the newtonian wall - bounded flow , making it more elaborate and more precise .
but the message of this letter will stay unchanged ; whatever is the model of choice , once endowed with effective viscosity @xmath30 instead of @xmath12 , there would exist a profile of @xmath30 that would result in a maximal possible velocity profile at the edge of existence of turbulent solutions .
that profile is the prediction of the said model of choice for the mdr .
in particular we offer a prediction for simulations : direct numerical simulations of the navier - stokes equations in a channel , endowed with a linear viscosity profile @xcite , will not be able to support turbulent solutions when the slope of the viscosity profile exceeds the critical value that is in correspondence with the slope of the mdr .
notwithstanding , it is gratifying to discover that even a simple model of the balance of energy and momentum is sufficient , in light of the insight presented in this letter , to predict ab - initio the functional form _ and _ the parameters that determine the maximum drag reduction asymptote . |
the dynamics of a charged particle in a strong magnetic field shows up a separation of time - scales , with the existence of a fast larmor rotation , slower drifts , and an adiabatic invariant .
this can be used to isolate completely the slow part of the dynamics from the single fast coordinate , the gyro - angle , and to build a true constant of motion , the magnetic moment @xcite .
thus the guiding - center reduction brings a slow reduced motion and decreases by two the effective dimension of the dynamics .
it is the starting point of gyrokinetics , which is a kinetic description of plasma dynamics in a strong magnetic field , and a key model in the study of plasma micro - turbulence @xcite .
the most efficient results for the guiding - center reduction are obtained by lie transforming the lagrangian @xmath0-form in an expansion in a small parameter related with the magnetic inhomogeneity at the scale of the larmor radius @xcite . in this paper , we clarify some aspects of guiding - center theory , especially some difficulties about the definition of the gyro - angle , by showing that the reduction can be done at arbitrary order in the larmor radius using gyro - gauge - independent coordinates .
+ indeed , the introduction of a scalar coordinate for the angle measuring the larmor rotation , namely the _ gyro - angle _ , imposes to choose at each point in the configuration space a direction , an axis , in the plane perpendicular to the magnetic field , defined as the zero from which the angle is measured .
this corresponds to fixing a gauge in the theory , the so - called gyro - gauge , and the presence of this gauge in the theory raised some non - trivial questions @xcite , even the bare existence of a global choice of gauge can fail .
so , it is interesting to consider the reduction from a more intrinsic point of view . in a previous work @xcite , we proposed a guiding - center reduction which avoided to introduce a gyro - gauge .
the idea was to lie transform directly the equations of motion instead of lie transforming the lagrangian , as is usually done @xcite .
this brought a much simplified reduction .
especially , it provided the minimal guiding - center reduction which concerned only four coordinates ( instead of six ) : the transformation generator had no gyro - angle component .
so , for the gyro - angle , the initial gauge - independent coordinate could be used , and no gauge fixing was needed .
this physical coordinate is the unit vector @xmath2 of the component of the velocity orthogonal to the magnetic field , which defines the direction of the perpendicular velocity .
a limitation of that approach was that it was not suited to non - minimal guiding - center reductions , for which the method relying on lie transforming the lagrangian appeared as necessary , or at least much more efficient .
for instance , it is interesting to have the slow reduced motion hamiltonian , but the hamiltonian structure of the reduced model is hard to deal with when working on the equations of motion .
in addition , the magnetic moment is usually taken as one of the reduced coordinates ; this can be done by working on the equations of motion , but it is not so straightforward @xcite . in a deeper way , the freedom involved in the gyro - averaged part of the coordinate change can be employed for the reduced motion to be as strongly reduced as possible , to make the guiding - center dynamics as simplified as possible .
obtaining such a maximal guiding - center reduction by lie transforming the equations of motion is far from simple , especially because it implies to solve non - trivial secular differential equations . on the contrary ,
lie transforming the lagrangian @xmath0-form basically relies on algebraic equations , and the requirements for a maximal reduction are not much more difficult to get than the minimal ones .
this method also guarantees that the 4-dimensional slow reduced motion is hamiltonian , by working on a quarter - canonical structure in the poisson bracket .
so , the goal is to use the physical gauge - independent coordinate also when lie transforming the lagrangian , in order to consider a gauge - independent maximal guiding - center reduction .
+ the introduction of a vectorial quantity @xmath2 for the gyro - angle coordinate raises some questions , because the coordinate system becomes constrained : the variable @xmath2 has to remain normalized and perpendicular to the magnetic field .
changing the spatial position @xmath3 implies to change the coordinate @xmath2 at the same time .
this induces a connection for a covariant derivative on a space - dependent circle , which is related to the circle - bundle picture underlying in the gyro - angle coordinate @xcite .
it was already present in @xcite when lie transforming the equations of motion , but it is more involved to deal with for the full guiding - center reduction , because the coordinate @xmath2 will be changed as well , and not only derivatives or vector fields are involved , but also differential forms .
the resulting reduction will naturally provide gauge - independent results , whereas in the usual approach , they were obtained only for a part of the reduced quantities .
this fact can shed interesting light on previous guiding - center results , especially those related to gauge invariance .
for instance , in the usual approach , the gradient is not gauge - independent , and the reduced poisson bracket involves a gauge - independent corrected gradient .
a comparison with the results of the gauge - independent formulation is a way to get an intrinsic interpretation for this corrected gradient . the results of the gauge - independent formulation
can also be used to explore other questions about the gauge - dependent approach , for instance related to gauge arbitrariness and anholonomy @xcite . to avoid confusion , they will be the topic of another paper @xcite . here
, we show how a maximal guiding - center reduction can be derived in a gauge - independent formulation to arbitrary order in the larmor radius .
the proof relies on explicit induction relations to all orders , because the induction can be written as a matrix product , with some coefficients being differential operators . through inversion of this matrix ,
a maximal reduction can be studied , towards a more complete reduction and a more general viewpoint on special reductions considered in previous works , such as the so - called hamiltonian and symplectic representations identified in @xcite . in the derivation
, the cotangent of the pitch - angle is used as the coordinate corresponding to the parallel velocity ( component of the velocity parallel to the magnetic field ) , since this coordinate simplified computations for the minimal guiding - center reduction in @xcite and made all formulae polynomials .
this will clarify why this polynomiality can be observed in the results of the full guiding - center reduction as well .
+ taking care of the reduction at higher order is interesting , if not needed , for two reasons . on the first hand , it is necessary to validate the gauge - independent approach .
indeed , the gauge issues became more sensitive when addressing the second - order reduction @xcite .
an acceptable solution to these issues , which is the goal of the present paper , needs to be appropriate for higher orders . on the other hand ,
it is motivated because standard works proceeded only up to part of the second - order reduction , but recent results emphasize the importance of higher - order terms , for instance because they are involved in the conservation of angular momentum @xcite and are crucial for a proper description of intrinsic rotation of tokamak plasmas , a key phenomenon to stabilize turbulence and increase the energy confinement time , which is the main goal of magnetic fusion .
+ the paper is organized as follows . in sect .
2 , a few facts are reminded about the initial dynamics , the choice of coordinates for the gyro - angle , the method of lie transforming the lagrangian @xmath0-form , and the hierarchy of requirements involved in the guiding - center reduction . for the sake of completeness and clarity , the mechanism of lie transforming the lagrangian through an expansion in a small parameter
is described in an appendix , with emphasis on the three steps it involves : an initialization for the lowest orders , whose choices are the key to make the reduction work and possibly be optimal ; an algorithm which applies for higher orders , is purely mechanical and can be applied to study the reduction to arbitrary order ; and an intermediate step in between . in sect .
3 , the method is applied to the guiding - center reduction in case the gyro - angle coordinate is chosen as the physical variable @xmath2 .
the derivation is written in matrix form , which emphasizes both the lowest - order choices that allow the reduction to work and to be maximal for the lagrangian , and the algorithmic character of the procedure at higher orders .
the full derivation is explained because we are interested in the reduction at arbitrary order in the larmor radius , which implies to use all the ingredients of the detailed mechanism at work . for orders lower than @xmath4
, the procedure follows the same lines as when working with the gauge - dependent coordinate , but formulae have to be used in their intrinsic version , for instance because the basis of @xmath0-forms involves non - closed @xmath0-forms .
finally , for orders higher than 2 , explicit formulae are given for the induction relations , allowing to go to arbitrary order , and to give a unified framework where recent results may seem to discord somehow with each other @xcite . in sect .
4 , the results are compared with previous works , either lie transforming the equations of motion , or using a gauge - dependent gyro - angle . for the sake of simplicity , we consider the special case where there is no electric field , but the generalization for a non - zero electric field is straightforward , as will be shown in subsect .
[ sectelecfield ] .
the dynamical system is simply a charged particle with position @xmath3 , momentum @xmath5 , mass @xmath6 and charge @xmath7 , under the influence of a static inhomogeneous magnetic field @xmath8 .
the motion is given by the lorentz force @xmath9 when the magnetic field is strong , the motion implies a separation of time - scales .
this is best seen by choosing convenient coordinates for the momentum space , for instance @xmath10 where @xmath11 is the unit vector of the magnetic field , and @xmath12 is the so - called _ perpendicular momentum _ , i.e. the orthogonal projection of the momentum onto the plane @xmath13 perpendicular to the magnetic field .
the coordinate @xmath14 is the norm of the momentum ; the coordinate @xmath15 is the so - called _ pitch - angle _ ,
i.e. the angle between the velocity and the magnetic field
. the last coordinate @xmath2 is the unit vector of the perpendicular velocity .
then , the equations of motion write @xmath16 where @xmath5 is now a shorthand for @xmath17 , the norm of the magnetic field @xmath18 is denoted by @xmath19 , and following littlejohn s notations @xcite , the vector @xmath20 is the unit vector of the larmor radius , so that @xmath21 is a right - handed orthonormal frame ( rotating with the momentum ) . in the case of a strong magnetic field ,
the only fast term is the larmor frequency @xmath22 .
writing the dynamics as @xmath23 , all other terms as @xmath24 appear to be of order @xmath25 , which means that the small parameter of the theory is of order @xmath26 . a more detailed study ( e.g. @xcite )
shows that it is rather @xmath27 it is an operator , but the gradient @xmath28 has only a meaning for orderings ; it acts on the magnetic field and can be given a more precise meaning by relations such as @xmath29 .
the ordering parameter @xmath30 is related to the magnetic inhomogeneity at the scale of the larmor radius @xmath31 . by abuse of language ,
it is often considered as just the larmor radius , or as the inverse charge @xmath32 @xcite .
the larmor frequency @xmath22 concerns only one coordinate , @xmath2 , the direction of the vector @xmath33 in the @xmath34-dimensional plane perpendicular to the magnetic field .
this corresponds to an angle , the so - called _ gyro - angle _ , and measures the larmor gyration of the particle momentum around the magnetic field . to get a true scalar angle instead of the vector @xmath2 ,
one chooses at each point @xmath3 in space a direction which will be considered as the reference axis @xmath35 .
this corresponds to fixing a gauge in the theory , the so - called _ gyro - gauge_. then , the gyro - angle @xmath36 is defined from the oriented angle between the chosen reference axis @xmath37 and the vector @xmath2 through the following relation : = - _ 1 - _ 2 .
[ defthetaperp ] the equation of motion for @xmath36 is @xmath38 with @xmath39 the unit vector such that @xmath40 is a ( fixed ) right - handed orthonormal frame @xcite .
+ the coordinate @xmath36 is not intrinsic , it depends on the chosen gauge @xmath37 , which raised some questions about the gauge - invariance of the theory , about the failure of global existence for @xmath41 , as well as about the presence of an anholonomic phase in the coordinate system @xcite .
all these difficulties originate because @xmath36 is not given by the physics , it is not a purely intrinsic coordinate . for physical results ,
what is needed is not @xmath36 but only @xmath2 .
this is clearly illustrated by all the results of guiding - center theory , e.g. @xcite , where @xmath36 intervenes only through the quantity @xmath2 everywhere ( except in its own definition and in subsequent relations ) .
so , we will avoid this coordinate and keep the corresponding initial variable , @xmath2 , as in @xcite .
the quantity @xmath36 will be used only with a symbolic meaning for the fast angle , or when making comparison with the gauge - dependent approach .
+ the use of a unit vector avoids having to fix a gauge for the zero of the angle , and it allows to work with a physical quantity : the unit vector of the perpendicular velocity : = = , [ defperp ] which indeed corresponds to the direction of the perpendicular velocity , and is a coordinate measuring the larmor gyration .
it is an angle , since it is a unit vector in a plane , namely the plane @xmath13 , orthogonal to the local magnetic field .
but this unit vector is immersed into @xmath42 , which means it is in @xmath43 .
this spatial dependence implies that the coordinate space is constrained : the gyro - angle @xmath2 is not independent of the spatial position .
when the coordinate @xmath3 is changed , the coordinate @xmath2 can not be kept unchanged , otherwise it may get out of @xmath13 .
differentiating relation ( [ defperp ] ) with respect to @xmath3 gives = - ( + ) .
[ connectnatur ] this formula can be obtained more easily by noticing that in the change of coordinates @xmath44 , the following relations hold @xmath45 in which @xmath28 means differentiation with respect to @xmath3 while keeping @xmath5 constant , and we used that @xmath46 and that @xmath47 is a unit vector , which implies @xmath48 .
formula ( [ connectnatur ] ) is not well defined where @xmath49 , i.e. where @xmath5 is parallel to @xmath8 . but
this is no trouble , since it fits in with a usual limitation of guiding - center theories .
for instance , guiding - center transformations are not defined where @xmath50 is zero , since they involve many @xmath51 as denominators @xcite . at the points where @xmath49 , the vector @xmath2 itself is not defined , neither is the coordinate @xmath36 .
it is an implicit assumption in all gyro - kinetics and guiding - center works that those points are excluded from the theory .
+ if the coordinate @xmath36 was used , formula ( [ connectnatur ] ) would be replaced by = - + _
1_2 , [ connectgauge ] where @xmath52 is the so - called _ gauge - vector _ , which is usually denoted by @xmath53 , depends only of the position @xmath3 , and is related to the choice of gauge . to fit in with both coordinates ,
we define a more general function @xmath54 : @xmath55 then , in any coordinates , the previous formulae ( [ connectnatur ] ) and ( [ connectgauge ] ) write = - + _ g .
[ connectboth ] the value @xmath56 corresponds to the physical definition of @xmath2 , and the corresponding formula ( [ connectnatur ] ) .
the value @xmath57 will be linked with the usual case relying on the gauge - dependent coordinate @xmath36 , according to formula ( [ connectgauge ] ) . the spatial dependence of the vector @xmath2 through formula ( [ connectboth ] )
must be taken into account each time a gradient acts on a function that depends on the fast angle @xmath2 , e.g. in total derivatives , in lie transforms , and when computing the action of a spatial component of the change of variables .
a more detailed study of the coordinate system @xcite shows that @xmath58 is the connection associated to the gyro - angle circle bundle , which indeed can be any function of the phase space , and that @xmath28 is the ( spatial part of the ) corresponding covariant derivative .
the goal is to isolate the slow part of the dynamics from the fast angle @xmath36 ( or @xmath2 ) , by performing a near - identity change of coordinates such that the dynamics of the remaining coordinates does not depend on @xmath36 ; this averaging procedure is the primary requirement for the guiding - center reduction .
we showed in @xcite how it can be derived by lie transforming the equations of motion when keeping the physical coordinate @xmath2 .
it appeared to be very straightforward for the minimal guiding - center reduction , but the additional requirements ( e.g. the use of the magnetic moment as a coordinate or a further simplification of the reduced dynamics ) were not so easy to obtain .
they are more efficiently obtained by lie transforming the lagrangian @xmath0-form , which relies on the hamiltonian structure of this dynamical system .
+ the corresponding hamiltonian function is just the particle kinetic energy @xmath59 the poisson bracket is non - canonical , it contains the gyro - magnetic coupling term \{f , g}= _ f _ g -_f_g - _ fe_g .
[ crochet ] together , the poisson bracket and the hamiltonian induce the equations of motion through hamilton s equations = \{,h } , where all the coordinates were grouped in a single phase - space vector @xmath60 . + instead of working with the poisson bracket and the hamiltonian , it is easier to work on the _ poincar - cartan @xmath0-form _ @xmath61 @xcite .
this last is usually called just a lagrangian @xmath0-form , or simply a _
lagrangian_. it concentrates all the information on the hamiltonian structure into one single quantity , which in addition is much less constrained than a poisson bracket .
it is a @xmath0-form , defined over a @xmath62-dimensional space @xmath63 by @xcite : = ( e+)d - h dt , [ lagrini ] which yields a variational formulation of the dynamics with the action @xcite : @xmath64 the symplectic part of @xmath61 is @xmath65 @xcite .
it is a @xmath0-form @xmath66 in the usual @xmath67-dimensional phase - space , which gives the lagrange @xmath34-form through exterior derivative _
s:=d_s=(_i_s^j-_j_s^i ) dz^i dz^j , [ extderiv ] where @xmath68 means tensorial product , and einstein s convention is used : there is an implicit summation over repeated indices . in turn , @xmath69 is invertible , and its inverse gives the poisson bracket @xmath70 with @xmath71 the bivector defined by the relation @xmath72 .
+ in the presence of a strong magnetic field , the small parameter @xmath30 allows for a perturbation expansion of all quantities . the lagrangian @xmath61 from formula ( [ lagrini ] )
can be written @xmath73 where the index refers to the order in the magnetic field , ( or in @xmath32 , following northrop s work @xcite ) _
-1&:=&ed , _ 0&:=&d-
dt , [ lagrordering ] whose ratio is indeed of order @xmath30 .
the lowest order being @xmath74 reminds that the guiding - center is a singular perturbation theory .
+ the lagrangian is transformed by the exponential of a lie transform @xmath75 where @xmath76 is the lie derivative along the vector field @xmath77 , whose inverse is the generator of the near - identity coordinate transformation : @xmath78 the generator can be expanded @xmath79 , where @xmath80 is the term of @xmath77 that is of order @xmath81 in @xmath30 .
in fact , it is equivalent but simpler to replace the single transformation @xmath82 with a complicated generator by a series of transformations with a simple generator for each of them @xmath83 where @xmath84 is the vector field generating the @xmath81-th transformation , and is purely of order @xmath85 .
the lagrangian then transforms as @xmath86 where @xmath87 is the lie derivative along the vector field @xmath88 .
notice that @xmath89 because @xmath90 .
+ as the lagrangian is time - independent , it is interesting to use time - independent perturbation theory , by imposing a transformation that does not depend on time and does not affect the time coordinate : @xmath91 is constant in time and @xmath92 .
this implies that the hamiltonian will transform as a scalar function .
indeed , for any time - independent vector field @xmath91 and any @xmath0-form , @xmath93_t+ [ d ( i_\bfmg\ga)]_t \ncr & = \left [ \bfmg^i\p_{\bfy^i}(\ga_j ) d\bfy^j - \bfmg^i\p_{\bfy^j}(\ga_i ) d\bfy^j \right]_t + \p_t(\bfmg\cdott\ga ) \ncr & = \bfmg\cdott\p_\bfy\ga_t + \p_t(\bfmg)\cdott\ga = \bfmg\cdott\p_\bfy\ga_t \ , , \notag\end{aligned}\ ] ] where formula ( [ liederiv ] ) was used for lie derivatives , and formula ( [ extderiv ] ) for exterior derivatives ( here in the @xmath62-dimensional space ) .
another consequence is that at each order @xmath81 in @xmath30 , there will be seven requirements ( see next subsection ) , one for each component of the reduced lagrangian @xmath94 , and only six freedoms , one for each component of the time - independent transformation generator @xmath88 .
one freedom is missing in @xmath88 and must be looked for elsewhere .
now , the lagrangian is defined only to within a total derivative , since only its exterior derivative has a physical meaning , and @xmath95 for any function @xmath96 , which is called a _ gauge function _ @xcite .
be careful , this gauge has nothing to see with the gyro - gauge nor the gyro - angle , it is just an arbitrariness in the definition of the lagrangian .
rather , it corresponds to the electromagnetic gauge , since it can be absorbed in a redefinition of the potential @xmath97 ( together with @xmath98 when there is a non - zero electric field , see page ): the lagrangian is expressed in terms of the potential , and does depend on the electromagnetic gauge , but the dynamics , as well as all physical quantities , are electromagnetic - gauge invariant , they depend only on the electromagnetic field . the freedom embodied in @xmath96 is needed to obtain the maximal reduction , since it gives the expected seventh freedom .
then the reduced lagrangian is : @xmath99 it will be determined order by order in @xmath85 : @xmath100\ga_{-1 } \ncr & \hspace{15ex } + \big[\sfml_3+\sfml_2\sfml_1+\tfrac{\sfml_1 ^ 3}{6}\big]\ga_0 + ds_3 \ncr ... & \notag\end{aligned}\ ] ] in principle , these are differential equations for @xmath91 , because the action of a lie derivative @xmath101 over a @xmath0-form @xmath102 writes _ = (
i_d + d i _ ) , [ liederiv ] where the operator @xmath103 is the interior product , e.g. i_= =( ) .
[ interiorproduct ] so , the first term in ( [ liederiv ] ) is algebraic in @xmath91 , but the second one is differential in @xmath91 .
however , the differential operators can be avoided by the following argument .
the last term in ( [ liederiv ] ) involves an exterior derivative , and can be removed by redefining the gauge function @xmath96 .
in addition , formula ( [ liederiv ] ) together with the property @xmath104 imply that the exterior derivative and the lie derivative commute : @xmath105 this means that for any vector fields @xmath77 and @xmath91 @xmath106 in computations for @xmath107 , the last term can again be removed by redefining the gauge function @xmath96 . by induction , it is now easy to see that in formula ( [ reduclagr ] ) , exponentials of lie derivatives can be considered as just exponentials of interior products provided the gauge function @xmath108 is defined in a convenient way at each order @xmath85 to absorb all the exterior derivatives involved in equation ( [ reduclagr ] ) : @xmath109 for the following , we will redefine @xmath96 according to this rule , but for simplicity , we drop the prime and write @xmath96 for @xmath110 . using this rule and the fact that in the equation for @xmath94 , the @xmath111 are already known , equations ( [ reduclagr ] ) can be written @xmath112 + ds_2 \ncr \overline \ga_{3 } & = \bfmg_4\cdott\om_{-1 } + \bfmg_3\cdott\overline\om_0 + \bfmg_2\cdott\overline\om_1 - \tfrac{(\bfmg_2\cdott d)^2}{2}\om_{-1 } \ncr & \hspace{20ex } + \tfrac{(\bfmg_1\cdott d)^2}{24 } \bfmg_1\cdott(3\om_0+\overline\om_0 ) + ds_3 \ncr ... & \notag\end{aligned}\ ] ] where the notation @xmath113 is used for the interior product , as in formula ( [ interiorproduct ] ) .
in addition , the @xmath81-@xmath114-order lagrange @xmath34-form was defined in the natural way : @xmath115 in the next subsection , we study the properties that are wished for @xmath107 . in the next section ,
the unknowns @xmath88 and @xmath108 will be determined such that @xmath94 has those desired properties .
\a ) the primary requirement for the guiding - center reduction is to isolate the slow dynamics of the coordinates @xmath116 from the fast gyro - angle @xmath117 .
from the point of view of the lagrangian @xmath107 , it may be obtained by making @xmath107 independent of @xmath117 .
this is actually stronger than the strict minimal requirement , since it implies to average the dynamics of @xmath117 as well .
however , in the lagrangian approach , contrary to when working on the equations of motion , the minimal requirement would be difficult to get , if not impossible , and it is quite easier to average all the reduced dynamics .
so , the goal is that the reduced lagrangian does not depend on the reduced gyro - angle , which means that all its non - zero fourier components ( i.e. purely oscillatory terms ) are zero : ( _ n ) = 0 , [ requirprior1 ] where following littlejohn s notations , @xmath118 is the projector onto gyro - fluctuations , with @xmath119 the complementary projector onto gyro - averages : @xmath120 for any function @xmath121 .
this average can be computed without introducing any gauge : either using the intrinsic calculus introduced in @xcite , either using the matrix calculus introduced in @xcite .
also , the coordinate @xmath36 can be used as an intermediate quantity for this computation , which is made at constant @xmath3 , so that the presence of a gauge ( only for the intermediate computation ) is of no consequence .
+ b ) averaging the motion or the lagrangian does not determine the average components of the coordinate change , as is clear in @xcite , for instance .
this lets some freedom in the procedure and suggests to impose stronger requirements for the reduction .
the basic idea is to use the available freedoms to make the reduced dynamics as simplified as possible .
a natural prospect is to make trivial the reduced dynamics @xmath122 for some components @xmath123 , by including constants of motion in the reduced coordinates . for the remaining coordinates
, one can consider putting their reduced dynamics to zero just for orders higher than @xmath34 or @xmath4 for instance : @xmath124 for all higher orders , where the index @xmath81 refers to the order in @xmath85 and the exponent @xmath123 indicates the component of the vector @xmath125 .
when this is achieved , the reduced dynamics is given just by lowest - order terms ; it is exactly known after the lowest orders have been derived , without computing the reduction at higher orders , which are useful only to determine the transformation . in the procedure
working on the lagrangian @xmath0-form , the `` components '' are not the ones of the reduced equations of motion , but the ones of the reduced lagrangian _
n^j=0 , [ requirprior3 ] for higher @xmath81 . for differential forms
, we use the same convention as for vectors : the index @xmath81 indicates the order in @xmath85 and the exponent @xmath123 refers to the component .
it departs from the usual notation @xmath126 , but it avoids excessive use of parentheses . equation ( [ requirprior3 ] ) gives additional ( i.e. non minimal ) requirements for the reduced lagrangian @xmath107 and it is used to determine the averaged transformation generators .
when it can not be obtained completely , the goal is to obtain it for as many components @xmath123 as possible , to get what can be considered as the maximal reduction .
+ c ) when studying the lie transform of the lagrangian @xmath0-form , one variable , namely the _ magnetic moment _
@xmath127 , plays a key role , as the variable conjugated to the gyro - angle . basically , including @xmath127 among the reduced coordinates is a way to obtain a more efficient reduction process by making one of the components of the reduced lagrangian trivial @xmath128 .
indeed , when the derivation is performed with the variable @xmath14 , then the @xmath129-component of the lagrangian is given by a whole series , which is the magnetic moment .
changing the variable @xmath14 in such a way that the new variable absorbs this series is a way to have the reduced @xmath129-component trivial , just given by a coordinate .
this also simplifies the reduction algorithm , by providing a simpler expression for @xmath130 , which will play a key role in the derivation .
+ second , the resulting variables @xmath131 and @xmath127 are conjugated , which implies that the magnetic moment is a constant of motion besides the norm of the momentum @xmath14 . and this conserved quantity is preferable to the variable @xmath14 because , unlike @xmath14 , it remains an adiabatic invariant in the presence of a wide class of electric fields .
the variable @xmath127 is a whole series in the larmor radius , and its lowest - order term is the well - known adiabatic invariant @xmath132 often confounded with @xmath127 : = .
[ magmomordmin ] so , an interesting additional requirement is to include the magnetic moment in the reduced coordinates . for the equation to be solved ,
this requirement is expressed by ^:= .
[ requirprior2 ] indeed , then the reduced poisson bracket verifies ^= 1 , ^i = 0 , [ qcancondition ] for @xmath133 , which means that the dynamics of @xmath127 is zero @xmath134 since the hamiltonian @xmath135 does not depend on @xmath117 , as a consequence of ( [ requirprior1 ] ) .
+ there is a third reason for including the magnetic moment among the reduced coordinates and requiring @xmath136 .
it is concerned with the hamiltonian structure of guiding - center dynamics .
the @xmath67-dimensional reduced motion @xmath137 is hamiltonian , since it is just the transform of the hamiltonian motion @xmath138 .
but the true reduced guiding - center motion is the @xmath139-dimensional slow motion @xmath140 .
it is the truncation of the full dynamics @xmath125 , but truncations of a hamiltonian dynamics are in general not hamiltonian .
however , in some cases , truncations are automatically hamiltonian , and a special case is the quarter canonical structure of the poisson bracket , defined by conditions ( [ qcancondition ] ) .
this can be seen by imposing dirac s constraints @xmath141 to the reduced dynamics , or by verifying that the truncated bracket is actually just given by starting from the initial lie algebra of all functions of the phase - space @xmath142 , and taking the sub - algebra of functions that do not depend on the gyro - angle @xmath143 .
thus , including @xmath127 in the reduced coordinates is a way to guarantee the reduced slow motion to be hamiltonian .
+ the requirement on the magnetic moment fixes one of the freedoms involved in the average components of the coordinate change .
the other freedoms are used to make the reduced dynamics as simple as possible , by putting to zero as many average components of @xmath94 as possible for higher @xmath81 , as indicated by formula ( [ requirprior3 ] ) .
+ d ) to sum it up , the guiding - center reduction involves a hierarchy of requirements : the primary requirement ( minimal reduction , with an averaged reduced dynamics ) is to remove the fast time - scale by averaging the lagrangian over the gyro - angle ; the corresponding equation is ( [ requirprior1 ] ) .
the secondary requirement ( intermediate reduction , with a constant of motion and a hamiltonian slow reduced motion ) is to include the magnetic moment among the reduced coordinates by the quarter - canonical structure ; the corresponding equation is ( [ requirprior2 ] ) . the third optional requirement ( maximal reduction , with a simplified reduced dynamics ) is to use the remaining freedoms to make the reduced lagrangian as simplified as possible ; the corresponding equation is ( [ requirprior3 ] ) .
this makes seven requirements at each order in @xmath30 , one for each component of the reduced lagrangian @xmath94 in formulae ( [ reduclagr ] ) , and seven freedoms are needed . for a time - independent transformation , those are @xmath144 and @xmath108 , as announced in the previous subsection .
let us now turn to the guiding - center reduction .
the details of the procedure as well as the practical computations may seem intricate and they hide somehow that the basic ideas of the reduction are very elementary .
it is why the principles and general lines of the procedure are presented in the appendix , to give a clear view of the reduction process . in this section
, the three stages of the method presented in the appendix are shown to work with the coordinate @xmath2 in a similar way as with the standard approach relying on a gyro - gauge .
the transformation at lowest orders is computed for comparison with previous works , and it is shown how the reduction can be performed to arbitrary order in the larmor radius by obtaining explicit induction relations .
each order of the derivation can be given several numbers .
for instance , what is usually called the first order is the order just after the lowest order . for the lagrangian , it corresponds to @xmath145 ( since the lowest order corresponds to @xmath146 ) , which is rather considered here as the order @xmath147 .
in addition , the order in the various quantities will be mixed up : for instance , in the derivation , the order involving @xmath148 will be the equation for @xmath149 , as well as for @xmath150 and also for @xmath151 . for the sake of clarity , we will always consider the order @xmath81 as the one corresponding to @xmath152 ( or rather @xmath94 ) , and we will often use the expression `` at order @xmath152 '' , instead of `` at the order corresponding to @xmath152 '' . the goal is to solve equation ( [ reduclagralg ] ) for the guiding - center reduction , with the lagrangian ( [ lagrordering ] ) , and with the requirements ( [ requirprior1])-([requirprior2 ] ) for the averaging reduction , for the magnetic moment reduction , and for the maximal reduction .
+ first of all , the change of coordinates from the norm of the momentum @xmath14 to the magnetic moment @xmath127 is not near identity , as is clear in equation ( [ magmomordmin ] ) . before beginning the reduction ,
a preliminary change of coordinates must be done , so that all the remaining transformation will be near identity .
a suitable preliminary change of coordinates is @xmath153 where @xmath132 is the zeroth - order magnetic moment @xmath154 which is a well - known adiabatic invariant , and often confounded with @xmath127 . in the new variables ,
the lagrangian becomes @xmath155 \cdott d\bfq - \mu b ( 1+\cot^2\varphi ) dt \,.\ ] ] interestingly , the pitch - angle @xmath50 intervenes only through its cotangent , which was mentioned in @xcite as making all quantities polynomials . here , this feature is obvious in the lagrangian , and it is preserved by derivatives , so that it will be preserved throughout all of the derivation . actually , the magnetic moment makes the polynomiality still more accurate than in @xcite , where the variables @xmath50 and @xmath14 were used , and the larmor - radius prefactor @xmath156 was not polynomial in @xmath157 . here , the magnetic moment @xmath132 absorbs the @xmath158 and all formulae will be purely polynomials in @xmath157 and monomials in @xmath159 , which is useful to simplify computations .
so , we actually choose to change coordinates according to @xmath160 with @xmath161 the variable that makes all formulae polynomials .
also , the structure of the lagrangian shows that one can make the coefficients @xmath7 and @xmath6 disappear by noticing that the magnetic field @xmath19 appears only through @xmath162 , provided @xmath132 is considered as appearing only through @xmath163 , and @xmath164 appears only through @xmath165 . the particle charge @xmath7 is usually kept in guiding - center works because the order in @xmath32 indicates the order in @xmath30 @xcite .
here , it is useless since the order in @xmath30 is already indicated by the order in other quantities : @xmath152 and @xmath94 are of order @xmath166 .
the reason is that all quantities will be series in @xmath167 , as a result of the structure of the lagrangian , and as will be confirmed by the derivation , e.g. formulae ( [ larmorradiusosc ] ) , ( [ resultsorderga11])-([resultsorderga12 ] ) , ( [ avgg1mu])-([hamiltredord2 ] ) , ( [ resultsdeparametrized1])-([resultsdeparametrized2 ] ) , etc .
hence the order can be readily obtained by the overall order in @xmath168 or @xmath28 .
these last two quantities have the drawback of being an operator , or a space - dependent function ; in addition , their order is only dimensional , e.g. @xmath169 is of order @xmath170 . for a readily control of the order , it is useful to have a scalar parameter , which was considered as @xmath32 in previous works . here , this role can be played by @xmath159 , since all quantities will be monomial in it .
thus , keeping @xmath7 to indicate the order of expansion is indeed not necessary .
thus , we make the scaling @xmath171 which avoids unnecessary coefficients in the derivation .
it agrees with the physics , where the effect of the magnetic field on particle dynamics always includes the coupling constant @xmath7 . for simplicity
, we will drop the underline , e.g. we will write @xmath19 for @xmath172 . the lagrangian becomes @xmath173\cdott d\bfq - \mu b ( 1+\phi^2 ) dt \,.\ ] ] the derivation starts from the lagrangian with the expansion ( [ lagrordering ] ) @xmath174 with @xmath175 it can be divided into its average and fluctuating part @xmath176 the previous section showed that the process involves the lagrange @xmath34-form @xmath177 .
it can be split in three basic terms @xmath178 with @xmath179 where the symbol @xmath180 denotes the antisymmetry operator @xmath181 for any matrix @xmath182 and any vectors @xmath183 and @xmath182 .
the first term in ( [ om3terms ] ) is the lowest - order lagrange @xmath34-form , related to larmor gyrations .
the second ( resp .
third ) term in ( [ om3terms ] ) is the exterior derivative of the oscillating ( resp .
average ) zeroth - order lagrangian .
be careful , this is not the oscillating ( resp .
average ) zeroth - order lagrange @xmath34-form ; for instance @xmath184 because the exterior derivative does not preserve gyro - fluctuations .
+ the contributions ( [ om3terms ] ) to the lagrange @xmath34-form are explicitly given by @xmath185 \phi\sfb \right\rbrace \cdott \wedge d\bfq \ncr & \hspace{10.5ex } - \big\lbrace ( 1+\phi^2)(b d\mu + \mu db ) + 2\mu b d\phi \big\rbrace \wedge \tfrac{dt}{m } \ncr \widetilde\om_0 : & = d \osc(\ga_0 ) = d ( \sqrt{2\mu b}\sfc ) \cdott\wedge d\bfq \ncr & = \sqrt{2\mu b } \left\lbrace d\sfc + \left [ \tfrac{d\bfq\cdott\na b}{2b } + \tfrac{d\mu}{2\mu } \right ] \sfc \right\rbrace \cdott \wedge d\bfq \ , .
\notag\end{aligned}\ ] ] now , @xmath186 involves two contributions : one corresponding purely to the gyro - angle , in which the variable @xmath2 is changed at constant @xmath3 , and a second one coming from a change in the coordinate @xmath3 , which comes because the gyro - angle is a constrained coordinate , as mentioned about formulae ( [ connectnatur ] ) and ( [ connectboth ] ) . the second part is given by @xmath187 the first one is then @xmath188 , but it is written more precisely as @xmath189 since when @xmath3 is constant , the variation of @xmath2 can only be in the direction of @xmath190 , i.e. along the circle @xmath191 . on the whole @xmath192 this shows that , in the gauge - independent approach , the natural form corresponding to @xmath1 is @xmath193 where the minus sign comes to agree with the usual convention ( [ defthetaperp ] ) on the orientation of the gyro - angle . unlike @xmath1 , this form is not closed in general ; it is why it is written @xmath194 . in the gauge - dependent case ,
it is closed and indeed corresponds to @xmath1 , as a consequence of ( [ connectgauge ] ) .
+ as the reduction process mainly relies on inversions of @xmath195 ( or rather @xmath196 ) , which can be viewed as a matrix inversion , a matrix notation is well suited and makes the discussion clearer .
this implies to choose a basis of @xmath0-forms ; the derivation of @xmath197 above shows that a natural basis is ( d , d , d | d , d , | -dt ) , [ basis1forms ] where a vertical dash @xmath198 is put to separate the space- , the momentum- and the time - components .
the choice of @xmath199 makes the corresponding coordinate @xmath200 instead of @xmath201 .
those @xmath0-forms are not closed , unlike the standard @xmath202 , but it is not needed , provided one is careful of using intrinsic definitions for the operations involved in the procedure ( see formulae ( [ basisvectfield ] ) and ( [ extderivgene ] ) for instance ) . in this basis , formulae ( [ formomegai ] ) become @xmath203[0ex]{\large 0 } & & \p_\mu h_0 \\ 0&0 & 0 & & & & 0 \\
\hline -\sfc\cdott\na h_0 & -\sfa\cdott\na h_0 & -\sfb\cdott\na h_0 & -\p_\phi h_0 & -\p_\mu h_0&0 & 0
\end{array } \right ) } \ncr \widetilde\om_0 & = ~~{\scriptsize \left ( \begin{array } { @{\hspace{1ex } } c@{\hspace{3ex } } c@{\hspace{4ex } } c@{\hspace{3ex } } | @{\hspace{5ex } } c@{\hspace{2ex } } cc@{\hspace{1ex } } | @{\hspace{3ex}}c@{\hspace{2ex } } } 0&-\tilde j&\tilde i & 0&-\tfrac{\sqrt{2\mu b}}{2\mu}&0 & \\
\tilde j&0&-\tilde k & 0&0&\sqrt{2\mu b } & \raisebox{-1ex}[0ex][0ex]{\large 0 } \\
-\tilde i&\tilde k&0 & 0&0&0 & \\ \hline 0&0&0 & & & & \\ \tfrac{\sqrt{2\mu b}}{2\mu}&0&0 & & \raisebox{-1ex}[0ex][0ex]{\large 0 } & & \raisebox{-1ex}[0ex][0ex]{\large 0 } \\
0&-\sqrt{2\mu b } & 0 & & & & \\ \hline & \raisebox{-1.2ex}[0ex][0ex]{\large 0 } & & & \raisebox{-1.2ex}[0ex][0ex]{\large 0 } & & \raisebox{-1.2ex}{\large 0 } \end{array } \right ) } \ , , \notag\end{aligned}\ ] ] where for visual purpose , vertical and horizontal lines are used to separate the position- , momentum- and time - components . the spatial part of the matrix @xmath204 ( resp .
@xmath205 ) is just a vector product @xmath206 , with the vector @xmath207 ( resp .
@xmath208 ) , where the coefficients are @xmath209 at lowest order @xmath210 , the reduced lagrangian ( [ reduclagralg ] ) writes @xmath211 the change of variable has no effect at this order , since it is near - identity .
but the averaging condition is trivially verified ; it is actually a condition for the near - identity lie transform to remove the fast time - scale , which is possible only when at lowest order , the lagrangian is already gyro - averaged .
now , the only freedom involved @xmath212 can not be useful and is set to zero @xmath213 at the following order @xmath214 , the reduced lagrangian is @xmath215 to write vectors in matrix form , a basis has to be chosen for vector fields also .
it is most convenient to choose the dual of the basis ( [ basis1forms ] ) for @xmath0-forms , so that the coupling @xmath216 is computed as a standard matrix product .
the desired basis is easily identified as ( using the natural isomorphism between vector fields and differential operators ) ( , , | _ , _ , - _
| -_t ) , [ basisvectfield ] because @xmath28 can be written @xmath217 .
the operator @xmath218 is the generator of larmor gyrations , as shown in @xcite .
it just an intrinsic definition of the usual @xmath219 .
the set ( [ basisvectfield ] ) is actually the natural basis for vector fields , which confirms the relevance of the basis ( [ basis1forms ] ) for @xmath0-forms .
be careful that the chosen bases are not averages and they must be taken into account when computing averages or fluctuations of a quantity .
for instance , it could seem that a vector ( resp .
a @xmath0-form ) with components @xmath220 is averaged , whereas it is not , since it is equal to @xmath221 ( resp .
@xmath222 ) .
+ then , equation ( [ lagrredordzeroeqz ] ) is easily computed in matrix form @xmath223 where , as usual , @xmath0-forms are written as @xmath224 matrices : each column is an equation to be solved for the freedoms @xmath225 and @xmath226 in such a way that @xmath227 satisfies the desired requirements .
the lowest - order lagrange @xmath34-form @xmath228 is linked to larmor gyration , as appears in ( [ formomegai ] ) .
it is not invertible .
only two components of @xmath225 are involved in the equation : @xmath229 and @xmath230 ; and only two components of @xmath107 can be controlled by these freedoms : @xmath231 and @xmath232 . as announced in the appendix ,
the inversion is possible only under some conditions on the term @xmath233 , and the solution , if it exists , is not unique . + more precisely , for the averaging requirement ( [ requirprior1 ] ) , the condition of the right - hand side of equation ( [ equag1matrix ] ) being in the range of @xmath228 is satisfied , since the only fluctuating terms are in @xmath234 , which can precisely be controlled by the freedoms @xmath235 .
the solution imposes the fluctuating part of @xmath235 , which is given by @xmath236 notice that the oscillating and averaged parts of the components @xmath229 and @xmath230 must be dealt with together through the combination @xmath237 .
this is easily illustrated with an example vector @xmath238 : it is not a pure fluctuation , whereas its @xmath2-component @xmath239 is a pure fluctuation , and its fluctuating part is @xmath240 , which mixes up the components @xmath241 and @xmath242 . at this point ,
the average part of @xmath234 remains free and must be identified by the secondary and tertiary requirements ( [ requirprior3 ] ) and ( [ requirprior2 ] ) . the magnetic moment requirement ( [ requirprior2 ] )
is not concerned here , since @xmath243 is automatically zero .
last , for the requirement for the maximal reduction ( [ requirprior3 ] ) , only the components @xmath234 can be controlled , and setting them to zero imposes the average part of @xmath235 @xmath244 formulae ( [ larmorradiusosc])-([larmorradiusavg ] ) give the traditional ( lowest - order ) larmor radius , which is usually confused with the exact larmor radius @xmath245 : @xmath246 where the index denotes the order , and formulae ( [ transfcoord ] ) and ( [ resultsorderga11 ] ) were implicitly used . in a similar way , what is usually called guiding center in the literature actually corresponds to the first - order guiding center : @xmath247 where we recover the traditional formula for the ( first - order ) guiding center , since we remind that here @xmath8 stands for @xmath248 ( see formula ( [ scaling ] ) ) . + all other components of the first - order transformation generator @xmath249 still remain undetermined .
they embody the non - uniqueness of the matrix inverse @xmath250 and the corresponding freedom will be useful for the solvability conditions at the next order . as a consequence , there will be some order mixing : the components @xmath251 of @xmath225 will be determined at higher order , at the same time as the components @xmath252 of @xmath253 .
last , the gauge function @xmath226 can not be useful and is set to zero @xmath254 the final zeroth - order reduced lagrangian ( [ lagrredordzeroeq ] ) writes @xmath255 it is just the average of the zeroth - order initial lagrangian @xmath145 : @xmath256 .
the zeroth - order reduced @xmath34-form @xmath204 is then d_0=_0 , which justifies the notation @xmath204 introduced in ( [ om3terms ] ) .
it is easy to see that the matrix @xmath257 is not invertible , for instance all its gyro - angle components are zero .
so , the pivotal matrix for high orders @xmath258 mentioned in the appendix ( see page ) is not identified yet . at the following order @xmath259
, the reduced lagrangian is given by ( [ reduclagralg ] ) , which is rather written with the unknown @xmath260 on the left - hand side @xmath261 as announced in the appendix , the pivotal matrix is the set of @xmath228 and @xmath262 , acting on the set of unknown components of @xmath263 .
again , it is not invertible ; this corresponds to the case where the inversion is possible only under some integrability conditions on the right - hand side @xmath264 , to which one must add the set of @xmath265 and @xmath266 acting on the set of components of @xmath263 that are already known , i.e. on @xmath235 . and
the solution , if it exists , is not unique .
the requirements concern the seven components of the reduced lagrangian @xmath283 , which are ideally put to zero .
the freedoms are embodied in six components of @xmath281 and in the gauge function @xmath108 .
more precisely , the primary requirement ( averaging reduction ) means that the fluctuating part of the reduced lagrangian must be zero : @xmath284 ; actually , the chosen basis for @xmath0-forms is not averaged , so that each component can not be averaged separately , e.g. neither @xmath285 nor @xmath286 are averages , but the sum of them @xmath287 is an average . the secondary requirement ( magnetic moment reduction )
means that the gyro - angle component of the lagrangian must be the magnetic moment : @xmath288 , and @xmath289 for @xmath290 .
the tertiary requirement ( maximal reduction ) means that the average part of the reduced lagrangian should be zero as well : @xmath291 ( except @xmath292 ) .
when integrability conditions can not be solved for so strong requirements , the tertiary requirement is released , but as little as possible , and one sets to zero as many components @xmath283 as possible .
+ let us solve equation ( [ lagrredord1 ] ) .
the pivotal matrix is not invertible .
it has six columns and seven rows ; the columns @xmath293 and @xmath132 are not linearly independent ; in addition , its column for @xmath129 is zero , which means that @xmath292 can not be controlled by the freedoms @xmath263 .
so , the existence of solution is submitted to integrability conditions ( by which the freedoms @xmath294 and @xmath295 can be constrained ) , and then the existing solutions are not unique .
+ the integrability conditions are automatically satisfied . the overall equation for the column @xmath293 will be automatically zero , because the procedure will imply @xmath296 , and @xmath297 . as for the column @xmath129
, it writes : @xmath298 the averaging requirement means that @xmath299 , which implies @xmath300 the average part of the equation then writes _ 1^= _
1^ = , [ magmomsucceeded ] which makes the automatically verified for this component . in fact , if @xmath132 had not been chosen as a preliminary coordinate , it is here that it could be identified . as a side comment
, it might seem that the right - hand side of equation ( [ magmomsucceeded ] ) should be @xmath127 instead of @xmath132 , because of formula ( [ requirprior2 ] ) .
however , in the derivation , all expressions for @xmath107 are functions of @xmath132 . in the reduced system
, the corresponding expressions will be the same functions evaluated on the reduced coordinate @xmath127 , as is well emphasized in @xcite .
so , formula ( [ magmomsucceeded ] ) means that the first - order reduced lagrangian @xmath301 will actually have its @xmath131 component equal to @xmath127 .
it is why this result verifies the requirement ( [ requirprior2 ] ) on the magnetic moment .
we will not insist more on this point . with the result ( [ magmomsucceeded ] ) ,
the reduced lagrangian will contain the term @xmath302 , implying the presence of the vector @xmath58 , which is not determined .
this feature already appeared in the computation of @xmath303 , but the vector @xmath58 involved in @xmath304 came from a term @xmath305 , so that the overall contribution of the vector @xmath58 actually cancelled .
just the same way now , the term @xmath306 is required to come from a total contribution @xmath307 this imposes a non - zero contribution for @xmath308 : @xmath309 for a maximal reduction , minimizing @xmath308 then means minimizing @xmath310 ( and ideally setting it to zero ) .
+ for the remaining @xmath311 columns , the pivotal @xmath312 matrix is invertible for the five unknowns @xmath313 , as is clear through the following argument : - the freedom @xmath277 controls the column for @xmath314 , since the coefficient @xmath315 is invertible and no other unknown component of @xmath263 appears in this column .
this does not determines fully @xmath277 , since the freedom @xmath316 appears in the column , and @xmath316 is to be identified by the column @xmath317 , in which @xmath277 appears again .
thus , the set of @xmath318 can be considered as a coupled set of equations for @xmath319 .
but it is solvable ; it implies that ( s_1 ) = k ( , ) , [ s1gene ] is an arbitrary function of @xmath320 and @xmath3 , and that _ 1^ = _ ( s_1 ) .
[ g2bgene ] - in a similar way , the set of @xmath321 is a coupled set of equations for @xmath322 , which is solvable . - indeed , the freedom @xmath323 controls the column for @xmath324 , since the coefficient @xmath325 is invertible .
the solution for @xmath323 is then parametrized by @xmath326 , which is still unknown , but appears in this column . -
then the freedom @xmath326 controls the column for @xmath327 , because when inserting the solution for @xmath323 , the coefficient of @xmath326 becomes just @xmath19 , which is invertible . -
the freedom @xmath328 controls the column for @xmath329 , since the coefficient @xmath330 is invertible and no other unknown component of @xmath263 appears in this column ( because now @xmath277 and @xmath326 are not unknowns any more ) . - the freedom @xmath331 controls the column for @xmath332 , since the coefficient @xmath19 is invertible .
the solution for @xmath331 is then parametrized by @xmath333 , which is still unknown , but appears in this column . -
then , all the components of @xmath334 remain free . they can be used to explore the various guiding - center representations @xcite . here , we are interested in a maximal reduction , which means to set them to zero , so that the optimal requirements are fulfilled . the freedom @xmath335 in @xmath316 ( which is a parameter in the formulae obtained for @xmath277 , @xmath323 , @xmath326 , @xmath328 , and @xmath331 ) can not be useful to improve anything and is set to zero : ( s_1)=0 .
[ avgs1settozero ] - the freedom @xmath333 is still undetermined , it embodies the non - uniqueness of the solution implied by the pivotal matrix being not invertible
. this freedom will be useful for the requirements at the following order . as a consequence
, the order mixing will not include only two orders , but three of them , and the pivotal matrix will act on components of @xmath336 .
in addition , the orders will not be solved independently , since the unknown @xmath333 will be identified at order @xmath337 , whereas it already appeared in the equations at order @xmath301 , so that it is a parameter in the expression computed for @xmath331 . +
this procedure shows that the average and fluctuating parts of the equations are dealt with the same way , because the equations for @xmath88 is algebraic .
this is very different from the minimal guiding - center reduction by lie transforming the equation of motion , whose equation relies on the operator @xmath219 , and which easily controls the fluctuating part of the equation , but involves secular differential equations for the average part of the equation @xcite .
+ at that point , it seems that the order @xmath301 has been completed : it is indeed completely satisfactory in itself , since all of the requirements are perfectly fulfilled , with the resulting reduced lagrangian @xmath338 .
however , the procedure will have to be slightly changed , because at the following order the secondary requirement for @xmath339 can be controlled by no higher - order freedom ; it can be controlled only by @xmath340 , which was already determined above by the tertiary requirement for @xmath341 .
these two requirements can not be simultaneously fulfilled and one of them has to be dropped .
it is here that the requirements are not dealt with in the same way : as one of them must be dropped , the choice is imposed by the hierarchy and the secondary requirement must be preferred to the tertiary one . as a consequence , @xmath340
must be let free at first order .
it remains a parameter in @xmath328 and in @xmath342 .
another consequence is that the tertiary requirement for @xmath341 has been lost , and @xmath342 has a non - zero value .
actually , the equation for @xmath342 was coupled with the one for @xmath324 , and the non - zero term can be put in either of these components of @xmath107 .
one can consider recovering a zero value for this term by using the freedom @xmath335 available in @xmath316 , which had been arbitrarily fixed to zero in the process above in equation ( [ avgs1settozero ] ) .
but the corresponding equation for @xmath335 has no solution .
indeed , requiring @xmath343 is an equation for @xmath335 , which is the only available freedom k = ( _ 1^ ) .
[ seculardiffequation ] with the expression ( [ avgg1mu ] ) for @xmath340 , it can be studied by expansion in series @xmath344 .
expanding the right- and left - hand side of the equation and equating the coefficient of the same orders in @xmath293 and @xmath132 gives a non - solvable equation .
for instance , the coefficient of order @xmath345 implies the following equation k_2 = - , [ obstruction ] which has no solution for a general magnetic geometry .
so , the only available freedom can not be used to obtain the full reduction @xmath346 and @xmath347 .
one of those components has to be non - zero .
+ computing explicitly the expressions for the solution at order @xmath301 according to the procedure identified above gives the following results
@xmath348 \ncr \osc(\bfmg_1^\mu ) & = \mu\tfrac{\sqrt{2\mu b}}{b } \big [ -\tfrac{b\apos \sfa}{b } -\phi\tfrac{\cba+\abc}{2 } -2\phi^2\abb \big ] \label{oscg1mu } \\
\avg(\bfmg_1^\phi ) & = \tfrac{1}{\sqrt{2\mu b } } \big [ \overline\ga_1'^\sfb + \mu \tfrac{\cba-\abc}{2 } \big ] -\tfrac{\phi}{2\mu}\avg(\bfmg_1^\mu ) \ncr \overline h_1 & = \sqrt{2\mu b}\phi \big [ \overline\ga_1'^\sfb + \mu\tfrac{\cba-\abc}{2 } \big ] + b\avg(\bfmg_1^\mu ) \ncr \bfmg_2^\sfa & = \tfrac{\mu}{b } \big [ \phi\tfrac{7\cba -9\abc}{4 } -\phi^2\abb \big ] + \tfrac{1}{2\sqrt{2\mu b}}\avg(\bfmg_1^\mu ) \ncr \bfmg_2^\sfc & = -\tfrac{\sqrt{2\mu b}}{b}\tfrac{\bfmg_1^\th}{2 } -\tfrac{\mu}{b}\aca \ , , \label{resultsorderga12}\end{aligned}\ ] ] where following littlejohn , a condensate notation is used for gradients : the curved prime is used to indicate gradients of the magnetic field , and the short overbar over a vector @xmath2 or @xmath190 indicates the matrix transpose ( for the euclidean scalar product ) , so that @xmath349 and @xmath350 .
this notation comes from @xcite , slightly adapted to make it more explicit , in order to fit in with higher - order expressions @xcite .
be careful , the straight prime and the long overline are different and do not indicate gradients or matrix transpose , e.g. in @xmath351 . finally , the first - order reduced lagrangian writes
_ 1 = -d + _ 1^d - h_1 dt .
[ lagrredord1sol ] in the results above , the term @xmath324 was kept free in order to include both of the choices considered above . as @xmath324 is free , the natural choice for a maximal reduction is @xmath346
; then the reduced hamiltonian is non - zero : @xmath352 alternatively , @xmath342 can be set to zero , by choosing _
1^ = - ( _ 1^ ) - .
[ choiceh1=0 ] this last choice is possible only if the inversion of @xmath293 does not cause a singularity , i.e. if @xmath340 has no overall contribution of order zero in @xmath293 , which will have to be verified when @xmath340 is identified ( see equation ( [ avgg1mu ] ) ) .
the first - order lagrangian ( [ lagrredord1sol ] ) induces the following first - order reduced lagrange matrix , which will be part of the pivotal matrix at higher orders : @xmath353[l]{\raisebox{0ex}[0pt][0pt]{\text{\large { $ \overline\om_1^{\bfq;\bfq}$ } } } } & & 0&\bfmr_g\cdott\sfa&0 & \p_a \overline h_1 \\ & & & - \p_\phi\overline\ga_1'^\sfb & - \p_\mu\overline\ga_1'^\sfb+\bfmr_g\cdott\sfb & 0 & \p_b \overline h_1 \\
\hline 0&0 & \p_\phi\overline\ga_1'^\sfb & 0&0&0 & \p_\phi \overline h_1 \\ -\bfmr_g\cdott\sfc&-\bfmr_g\cdott\sfa & -\bfmr_g\cdott\sfb+\p_\mu\overline\ga_1'^\sfb & 0&0&1 & \p_\mu \overline h_1 \\ 0&0&0 & 0&-1&0 & 0 \\ \hline -\p_c \overline h_1 & -\p_a \overline h_1 & -\p_b \overline h_1 & -\p_\phi \overline h_1 & -\p_\mu \overline h_1 & 0 & 0 \end{array } \right ) } \ , , \label{baromega1}\end{aligned}\ ] ] where the matrix @xmath354 is defined by @xmath355 d\bfq \ , .
\notag\end{aligned}\ ] ] at the end of the first - order analysis , three freedoms remain : @xmath333 , which is a parameter in @xmath331 ; @xmath326 , which is a parameter in @xmath323 , in @xmath328 and in @xmath342 ; and either @xmath324 or @xmath342 , which is a parameter in the other one .
+ the results ( [ resultsorderga11])-([resultsorderga12 ] ) have physical implications . for instance , @xmath356 determines ( together with @xmath225 ) the perpendicular component of the second - order larmor radius @xmath357 .
its averaged contribution will imply that the larmor radius is not a pure fluctuation , and later on , a non - zero @xmath358 will be obtained , which will imply that the larmor radius is not purely transverse to the magnetic field .
those features are well known in guiding - center works , and we do not insist on them . here , we focus on the mechanism of the reduction , to show how it can be performed to arbitrary order in the larmor radius using gauge - independent coordinates for the gyro - angle , and why it can be considered as a maximal reduction .
+ this point is the turning point of all the reduction : the matrix @xmath359 is invertible .
thus , in the induction for high orders , the pivotal matrix @xmath258 will be the set of @xmath360 acting on the set of unknown components of @xmath361 .
it means that with respect to the reduction procedure announced in [ sect : appendi ] , the first stage of the reduction is achieved , and the order at which @xmath362 becomes invertible is @xmath363 .
so , the critical order at which the pivotal matrix becomes the same at each order is @xmath364 ( see the appendix for a more detailed analysis of the reduction procedure , with especially the role of the orders @xmath365 and @xmath366 , introduced in pages and ) .
accordingly , the algorithm for the derivation at high orders ( third stage mentioned in the appendix ) can be identified by now , but it will be efficient only for orders @xmath367 . the intermediate orders ( named `` second stage '' in the appendix ) must be studied separately . in order to introduce the derivation order by order
, we first go through the second stage and postpone the third stage , but it is important to notice that this last is independent of the second stage and could be studied before .
especially , all the high - orders algorithm relies on the matrix @xmath258 , together with the differential operators involved in @xmath368 .
they are already known by now , and are determined by the choices that have been made previously , and mainly by the choices at order @xmath0 . at the following order @xmath369
, the reduced lagrangian is given by ( [ reduclagralg ] ) _ 2 = _
3_-1 + _ 2_0 + d + ds_2 .
[ lagrredord2z ] at the previous orders , it appeared that for the algebraic part of the equations , the requirements on the average lagrangian were dealt with exactly the same way as the requirements on the fluctuating lagrangian .
so , they will not be studied separately .
only the integrability condition will restore a difference between them for some of the unknowns . at this point ,
the scheme for the unknown @xmath370 is not purely algebraical : @xmath225 is still not completely known , and it is involved in a first - order differential non - linear equation because of the term @xmath371 , with @xmath372 in equation ( [ lagrredord2z ] ) , the exterior derivative must be computed for the @xmath0-form @xmath373 that is not explicitly known yet . unlike in previous subsections , an explicit computation is not possible , and an abstract formula must be used ; care must be taken that the usual formula ( [ extderiv ] ) for exterior derivative can not be used , because it is valid only when the basis is composed of closed @xmath0-forms .
otherwise , it is replaced by the more general formula : @xmath374 for any @xmath0-form @xmath102 , with components @xmath375 in the basis @xmath376 . in formula ( [ extderivgene ] ) ,
two operations are to be identified : the action of the exterior derivative on scalar functions @xmath377 expressed in the chosen basis for @xmath0-forms , and the exterior derivative of the basis @xmath378 . for the exterior derivative on scalar functions @xmath377 , it writes as usual
@xmath379 provided the differential operators @xmath380 are given by the dual basis ( [ basisvectfield ] ) to the chosen basis ( [ basis1forms ] ) of @xmath0 forms @xmath381 . as for the exterior derivatives of the basis @xmath382 , they are easily computed as : @xmath383 in @xmath384 , the derivative @xmath385 appears .
such terms should be avoided , since they are not determined , as shown by ( [ connectboth ] ) .
but there is no trouble here , since in all the @xmath0-forms in this derivation , the term @xmath386 comes from @xmath387 .
the exterior derivative of the second term will generates @xmath388 , which will automatically cancel the term @xmath58 coming from @xmath389 . as a result
, there is no need to compute those terms .
computing the exterior derivative of a @xmath0-form @xmath102 can be made according to the following procedure : write @xmath390 , where now the spatial components @xmath391 do not involve @xmath392 . then apply formula ( [ extderivgene ] ) to @xmath393 and to @xmath394 .
the cancellation of the terms containing @xmath395 gives the resulting formula : d = _
i(_i^j-_j^i)_j + ^i d_i - _ i_i^_gd , [ extderivpractical ] in which @xmath396 , where @xmath397 is the kronecker delta , which means that @xmath398 is exactly @xmath382 but without the problematic term in @xmath399 .
+ with formula ( [ extderivpractical ] ) , the quadratic term in @xmath225 is found to have its momentum components linear in the unknowns @xmath400 , and differential only for @xmath333 : @xmath401 \label{quadraticg1momentum } \\ 6 \big ( i_{\bfmg_1}d\lambda \big)^\mu & = -\bfmg_1^\sfa \p_\mu \big [ 2\bfmg_1^\sfa\bfmr_g\cdott\sfa - 2\bfmg_1^\th \big ] + 4\bfmg_1^\sfa\bfmr_g\cdott\sfa - 4 \bfmg_1^\th \ncr 6 \big ( i_{\bfmg_1}d\lambda \big)^\th & = -\bfmg_1^\sfa
\p_\th \big [ 2\bfmg_1^\sfa\bfmr_g\cdott\sfa - 2\bfmg_1^\th \big ] + 6\bfmg_1^\mu + \bfmg_1^\sfa .6 j_{322 } \ , , \notag\end{aligned}\ ] ] where the coefficient @xmath402 is defined by @xmath403 \,.\ ] ] notice that it does not depend on @xmath58 , precisely because the term @xmath404 coming from formula ( [ extderivgene ] ) cancels the corresponding term in @xmath405 . + the differential equation for @xmath333 does not make things much more complicated , since it can be easily solved , e.g. by expansion in @xmath159 and @xmath293 . alternatively , a trick can be used to make the scheme purely algebraical @xcite ; it is not essential for the derivation , but we will use it because it simplifies much the explanations .
the idea is to notice that in this case , the only differential operators involved come from the second term of the exterior derivative @xmath406 , i.e. they are involved in expressions that write @xmath407 . in addition , in the sum over the index @xmath408 , only one of the terms involves a differential operator @xmath409 acting on the unknown @xmath333 , namely the term with @xmath410 . an integration by parts over this term can transfer the differential operator over the pre - factor @xmath230 , which is already known . this is a way to make the equation algebraic .
this integration by parts is justified by the gauge function . the equation for @xmath337
can be added a total derivative , which can be chosen @xmath411 and extracted from @xmath412 by @xmath413 : @xmath414 and this formula is not differential any more , but algebraic for @xmath415 , which is contained in @xmath416 .
+ then , equation ( [ lagrredord2z ] ) becomes _ 2 = _
3_-1 + _ 2_0 + + ds_2 , [ lagrredord2zz ] with @xmath417[0ex]{\large 0}&~~ & 0&0&1 & \raisebox{-1ex}[0ex][0ex]{\large 0 } \\ & & & 0&-1&0 & \end{array } \right ) } \notag \,,\end{aligned}\ ] ] where the momentum components are written in matrix form because they will determine the unknowns @xmath418 .
now , the induction relation just relies on a pivotal matrix @xmath419 , which appears in the set of @xmath420 in which a linear algebraic operator ( a matrix ) acts on the unknown components of @xmath421 .
it is invertible in the sense that it determines six unknown components of @xmath421 , which is the maximum that can be done at each order .
remember the matrix can not be fully invertible , since the transformation is time - independent , so that @xmath88 is @xmath67-dimensional , whereas the matrix has value in the @xmath62-dimensional space @xmath422 ; and by the way , the @xmath62-dimensional matrix is anti - symmetric , hence not invertible .
the seventh requirement is to be provided by the gauge function @xmath423 , which is the only integrability condition involved at this order . to solve the equation , an analysis similar to the one at the previous order leads to the procedure summarized in the following tabular , where each row corresponds to one of the equations .
the component @xmath426 of the reduced lagrangian involved in the corresponding equation is indicated in the first column ; the unknown which controls the equation and permits @xmath427 is indicated in the second column ; the coefficient to be inverted is indicated in the third column .
@xmath428 a few comments are in place .
as had been announced in the derivation of formula ( [ oscg1mu ] ) , the new feature is that the equation for @xmath429 can be controlled only by the average first - order magnetic moment @xmath340 .
it is why it was not available at the previous order .
the coefficient @xmath219 is an operator , but it is invertible over gyro - fluctuations .
the set of @xmath430 is a coupled set of equations for @xmath431 , but each of the unknowns can be assigned to one of the equation because the system can be solved in the following way . the freedom @xmath432 controls the column for @xmath433 , since the coefficient @xmath325 is invertible .
the solution for @xmath432 is parametrized by @xmath434 , which is still unknown , but appears in this equation .
then , the freedom @xmath434 controls the column for @xmath435 , because when inserting the solution for @xmath432 , the coefficient of @xmath434 becomes @xmath19 , which is invertible . at the end ,
the reduced lagrangian @xmath337 is free and can be set to zero , as required for the maximal reduction .
the freedom @xmath436 ( which is a parameter in the formulae obtained for @xmath333 , @xmath358 , @xmath432 , @xmath434 , @xmath437 , and @xmath438 ) can not be useful and is set to zero .
+ in a similar way as at the previous order , at that point , the order @xmath337 is completely satisfactory in itself , since the reduced lagrangian has been fully reduced @xmath439 .
however , the procedure will have to be changed , because at the following order the secondary requirement for @xmath440 can be controlled only by @xmath441 , which is therefore not available to get the tertiary requirement for @xmath442 .
so , this last requirement has to be dropped .
accordingly , @xmath441 remains free at this order , and it is a parameter in @xmath437 and in @xmath442 .
this last has a non - zero value , but its equation was coupled with the one for @xmath433 , and the non - zero term can be put in either of these components of @xmath337 .
+ one can consider recovering a zero value for this term by using the freedom @xmath436 , whose value had been arbitrarily fixed to zero in the process above . indeed ,
if @xmath443 is let as a free parameter in @xmath358 , @xmath333 and @xmath432 , then when replacing these variables by their expression , the requirement @xmath444 becomes an equation for @xmath436 .
unfortunately , this equation is not easily studied . as a first attempt
, @xmath423 can be chosen such that it is absent from @xmath333 , as it was done in equations ( [ s1gene])-([g2bgene ] ) .
then it is easy to see that the equation for @xmath436 will have the same structure as equation ( [ seculardiffequation ] ) .
this equation may not be integrable , as it was the case for equation ( [ seculardiffequation ] ) . in this case
, one should relax the condition for @xmath436 to be absent from @xmath333 , and the differential equation for @xmath436 could be more difficult to study , because @xmath333 is involved in the @xmath0-form @xmath445 in equation ( [ lagrredord2 ] ) in a rather intricate way .
+ applying the procedure identified above gives @xmath446 where @xmath447 with the results ( [ avgg1mu])-([hamiltredord2 ] ) , the parameters involved in formulae ( [ resultsorderga11])-([resultsorderga12 ] ) can be made explicit : @xmath448 \label{resultsdeparametrized1 } \\
\overline h_1 & = \sqrt{2\mu b}\phi \big [ \overline\ga_1^\sfb -\mu\tfrac{\cba-\abc}{2 } \big ] \ncr \bfmg_2^\sfa & = \tfrac{\mu}{b } \big [ \phi\tfrac{5\cba -7\abc}{4 } -\phi^2\abb \big ] \ncr \bfmg_2^\sfc & = \tfrac{\mu}{b } \big [ -\tfrac{b\apos \sfc}{b } + \phi\tfrac{\aba-\cbc}{4 } - \phi^2\cbb \big ] + \tfrac{\sqrt{2\mu b } } { b } \tfrac{\p_\mu \avg(s_2')}{2 } \ , .
\label{resultsdeparametrized2}\end{aligned}\ ] ] the reduced first - order hamiltonian for the choice @xmath449 is @xmath450 the reverse choice @xmath347 is possible , since its existence condition in equation ( [ choiceh1=0 ] ) is satisfied , as is clear in ( [ avgg1mu ] ) .
it corresponds to a component @xmath451 of the reduced lagrangian given by _ 1^= , [ lagrredord1alain ] which is regular in @xmath452 , as expected .
in formulae ( [ avgg1mu])-([resultsdeparametrized2 ] ) , the components @xmath225 and @xmath453 have been completely computed and simplified for comparison with previous results , because it is the point where usual derivations stop . for the other components
@xmath432 , @xmath438 , @xmath437 , and @xmath454 , formulae ( [ avgg1mu])-([resultsdeparametrized2 ] ) are explicit solutions .
their right - hand side involves only known quantities ( or quantities that are free parameters for these relations ) , but it has not been not expanded and simplified .
this can can be done in a straightforward way just by computing explicitly the terms involved , but we will not pursue in that direction , since the calculation for the @xmath455 is lengthy , and useless for our purpose , which is just to show how the procedure can be performed to arbitrary order .
in addition , they are the topic of a work by the authors of @xcite , and were already partly introduced in @xcite .
+ in the results , the term @xmath433 was kept free to include both of the choices considered above : setting the non - zero term in @xmath456 just means choosing @xmath449 .
the other choice @xmath457 corresponds to @xmath458 this last choice is possible only if the term inside the curled parentheses has no overall contribution of order zero in @xmath293 .
also @xmath436 was kept free , because one can consider using it to obtain the full reduction with both @xmath449 and @xmath457 : it will imply a differential equation for @xmath436 , which might not be integrable , just as happened at the previous order in equation ( [ seculardiffequation ] ) , but it might be integrable , or partly integrable , and then provide the full reduction , or at least a stronger reduction .
otherwise , it can be set to zero . at the end of the analysis at order @xmath369 ,
it remains one unknown , one freedom , and a binary choice .
the unknown is @xmath441 ; it is a parameter in @xmath432 and in @xmath456 ( or @xmath433 ) ; it will be determined at the following order .
the freedom is @xmath436 , which remains free , but could not be used to improve the reduction .
the binary choice is that either @xmath433 or @xmath456 is set to zero , the other is computed accordingly .
this is very similar to what had occurred at the previous order , but now , the unknown @xmath459 does not appear as a parameter , since it is not at all involved at this order , as is clear in equation ( [ lagrredord2 ] ) .
+ let us turn now to the following order @xmath460 .
the equation writes ( [ reduclagralg ] ) @xmath461 the unknowns are components of @xmath462 .
the pivotal matrix is not @xmath258 , i.e. not just given by @xmath463 , because of the correcting term @xmath464 .
this term might make the derivation more difficult , since it is not algebraic , but non - linear and differential for @xmath253 @xmath465 \,.\ ] ] it can be written in matrix form @xmath466 where the matrices @xmath467 have obvious definitions , and are independent of @xmath468 .
this is enough to show that this additional term can be transferred into the right - hand side , i.e. it involves only terms that are already known at each step of the computation . when computing @xmath469 for the unknowns @xmath459 , @xmath441 , @xmath470 and @xmath471 , the only components of @xmath253 involved in the correcting term are in @xmath453 , which is already known at that point .
then , when computing @xmath472 for the unknowns @xmath473 , @xmath474 , @xmath475 , and @xmath476 all the components of @xmath253 are involved in the correcting term , but they are all known at that point . as a consequence , the correcting term can be put in the right - hand side of the equation , and the pivotal matrix is actually @xmath258 , i.e. the set of @xmath477 acting on the unknown components of @xmath462 .
this means that the critical order at which the pivotal matrix becomes the same at each order is @xmath478 , and the order @xmath460 can be included in the third stage , with all higher orders , which is studied in the following section .
now , the second stage of the method mentioned in appendix is ended and the third stage is beginning , which means that the matrix to be inverted is always the same at any order @xmath479 , and it is indeed invertible .
so , the reduction can be performed to arbitrary order .
the only possible complication comes from the integrability condition for the gauge - function @xmath108 ( and possibly @xmath94 ) , but after settling it , the reduction process becomes fully algorithmic and unique .
this is proven by induction .
let us suppose that at some order @xmath479 , the set of unknowns are @xmath480 which means that before that order , all lower - order quantities @xmath481 are already determined , and that after that order , all higher - order quantities @xmath482 will remain free parameters .
this assumption is verified at order @xmath483 , which initializes the induction .
as announced in the appendix , we have included the reduced lagrangian @xmath94 in the vector @xmath484 , because some components of @xmath94 can not be set to zero and have to be computed in the process .
the reduced lagrangian is given by equation ( [ matrpivot ] ) @xmath485 where @xmath486 indicates all terms of @xmath487 that are of order @xmath81 but do not involve @xmath488 , @xmath489 or @xmath490 : @xmath491_n - [ \bfmg_{n+1}\cdott\om_{-1 } + \bfmg_n\cdott\overline\om_0 + \bfmg_{n-1}\cdott\overline\om_1 ] \,,\ ] ] in which the index @xmath81 indicates the term of order @xmath81 .
+ then , the pivotal matrix @xmath258 is the set of @xmath492 acting on the unknown components of @xmath361 .
it is invertible in the sense that it determines six unknown components of @xmath361 , which is the maximum that can be done at each order .
the last requirement is provided by the integrability condition for the gauge function @xmath108 .
more precisely , grouping as usual in the left - hand side only the terms with unknown components of @xmath361 , and removing all the coefficients that do not contribute , the induction relation for @xmath94 writes @xmath493 where the terms involving known components of @xmath361 have been grouped with @xmath486 : @xmath494[l]{\raisebox{0ex}[0pt][0pt]{\text{\large { $ \overline\om_1^{\bfq;\bfq}$ } } } } & & 0&0&0 & \p_a \overline h_1 \\ & & & - \p_\phi\overline\ga_1'^\sfb & - \p_\mu\overline\ga_1'^\sfb & 0 & \p_b \overline h_1 \\
\hline 0&0 & \p_\phi\overline\ga_1'^\sfb & 0&0&0 & \p_\phi \overline h_1 \\ 0&0 & \p_\mu\overline\ga_1'^\sfb & 0&0&1 & \p_\mu \overline h_1 \\ 0&0&0 & 0&-1&0 & 0 \\
\hline -\p_c \overline h_1 & -\p_a \overline h_1 & -\p_b \overline h_1 & -\p_\phi \overline h_1 & -\p_\mu \overline h_1 & 0 & 0 \end{array } \right ) } \ , . \notag\end{aligned}\ ] ] again , the same comments as the ones after equation ( [ lagrredord1 ] ) are in place .
also , when computing @xmath486 , formula ( [ extderivpractical ] ) is to be used to account for the derivative of the chosen basis of @xmath0-forms and for the cancellation of the derivatives of @xmath58 .
+ when solving the induction relation , the mechanism is the same as at order @xmath227 for @xmath495 , and the same as at order @xmath301 for @xmath496 and @xmath497 .
in addition , the new feature is the presence of @xmath498 for @xmath499 and @xmath500 , but the mechanism is similar to what happens at order @xmath337 , in the sense that the pivotal coefficients are the same .
mainly , the procedure relies on three conjugation - like relations : @xmath501 are conjugated for @xmath228 ; @xmath502 are half - conjugated for @xmath503 , in the sense that the structure is quarter - canonical ; and @xmath504 are half - conjugated for the symplectic part of @xmath505 .
the same procedure can be applied , which is reminded in the following tabular .
@xmath506 then , all the components of @xmath152 remain free .
this can be used to explore the various guiding - center representations at higher orders @xcite . here
, we are interested in a maximal reduction , which means to set them to zero , so that the optimal requirements are fulfilled . the freedom @xmath443 , which is a parameter in the formulae obtained for @xmath507 , @xmath508 , @xmath509 , @xmath510 , @xmath511 , and @xmath512 , can not be useful and can be set to zero .
+ in the same way as at orders @xmath0 and @xmath34 , @xmath513 must remain free at that order , because it will be needed to solve @xmath514 at the following order , just as @xmath515 is needed here to solve @xmath516 .
so , one can not have the reduced lagrangian fully simplified @xmath517 .
one of its component remains uncontrolled , either @xmath518 or @xmath519 .
one can consider recovering a zero value for this component by using the freedom @xmath443 , whose value had been arbitrarily fixed to zero in the process above .
then , when computing @xmath507 , @xmath508 and @xmath509 , the average gauge function @xmath443 remains a free parameter .
when replacing these variables by their expressions , the equation @xmath520 becomes a differential equation for @xmath443 , whose structure is ( s_n ) = o.t .
, [ seculareqordn ] where @xmath521 means other terms that can be explicitly computed .
this resembles equation ( [ seculardiffequation ] ) , but here , the condition ( [ s1gene ] ) has not been required in the process , contrary to what happened in previous orders .
the reason is that now @xmath522 is controlled by @xmath507 , which has no effect on @xmath523 .
the integrability of equation ( [ seculareqordn ] ) will depend on the right - hand side and must be studied at each order ; a priori , it is not guaranteed , since obstructions such as ( [ obstruction ] ) are possible .
so , a systematic procedure can not use the freedom @xmath443 to get the additional requirement @xmath524 , which must be dropped .
then the freedom @xmath443 is useless and can be set to zero .
+ at the end of the @xmath81-th - order analysis , exactly all of the unknowns @xmath484 have been determined .
all the components of @xmath361 that remain unknown are in @xmath482 . yet
, this does not allow us to conclude that the induction is proven , because the unknown @xmath525 already appeared as a parameter in @xmath511 and in @xmath342 ; hence it is not completely free , whereas the induction assumes it is free ( independent of the quantities @xmath526 ) , since it is in @xmath527 ; it will be determined at the next order , and this could imply coupled equations , whose solvability is to be verified .
however , @xmath525 will be computed in equation for @xmath528 , which corresponds to the column @xmath529 in equation ( [ lagrredordn ] ) at the next order , and does not involve any of the parameter - dependent quantities . thus , there is no coupled equations , and the solutions are indeed explicit .
this terminates the proof of the induction : the reduction can be performed to arbitrary order in @xmath30 . notice that here
, the induction relation is considered from the point of view of @xmath94 ; this caused an interlocking between the orders , where in the solution at each order , a parameter is involved , which will be identified at the next order , when computing @xmath530 . to avoid this interlocking phenomenon , it is possible to consider the induction relation from the point of view of @xmath531 . the drawback would be that when solving the equations for @xmath530 , one would begin the heavy computations for @xmath532 , which are involved in @xmath533 , hence at the next order . in computations by hands
, this can be a trouble , but when using computer - assisted computations , this is no trouble and it would probably be a more relevant choice .
the argument above emphasizes the distinction to be made between four kinds of quantities .
first , some of the quantities are already known before the computation at order @xmath81 , namely @xmath534 .
second , for @xmath535 the components @xmath283 have not been computed yet , but they can be excluded both from the unknowns and from the parameters , since the algorithm shows that before any computation , they are already known to be zero for all @xmath536 , to fulfil the requirements ( [ requirprior1])-([requirprior2 ] ) for @xmath107 .
third , some quantities are not known yet , and will be determined after the matrix inversion , namely @xmath537 in which a vertical dash @xmath198 was written at the places where a vertical line will be written in the matrix @xmath258 below .
last , other quantities are not known and will remain free after the matrix inversion , namely @xmath519 , @xmath443 and @xmath538 ; the variables @xmath538 will be determined at higher order , but one of its component , @xmath525 , is already involved in the equations at order @xmath81 and behaves as a parameter in this matrix inversion .
so , the parameters are @xmath539 notice that @xmath540 is included in the parameters @xmath541 even if it is not an element of @xmath484 but of @xmath527 .
+ with the procedure above , the left - hand side of equation ( [ lagrredordn ] ) can be written as just a matrix product , provided the pivotal matrix is extended , to act on all the quantities @xmath542 to be computed at this order , even the reduced hamiltonian @xmath523 and the gauge function @xmath108 . to include the gauge function @xmath108 in the vector which is acted upon by the matrix ,
some coefficients in the matrix must be operators , and the equation will be transposed , so that the operators act on their right . for clarity , the order of the columns is chosen to fit with the steps of the algorithm @xmath543 so that the equations are solved one after the other in order .
a vertical dash @xmath198 was written at the places where a horizontal line will be written in the matrix @xmath258 below . with this order for the rows and for the columns ,
the equation becomes _ ( _ n)_^t + _ ( _ n)_^t + ^t=0 .
[ inductionordn ] here , the rows are the equations to be solved , corresponding to the ( re - ordered ) columns of equation ( [ lagrredordn ] ) .
the first term involves exactly the unknown quantities to be identified at this order .
the second term involves exactly the parameters involved at this order but which will remain free at the end of this order .
the third term involves only quantities that are already known at the beginning of this order .
the matrices are given by _ : = ( c@ c | cccc | cccc 1 & & & & & & & & & + & _ & & & & & & & & + & _ & - & & & & & & & + & _ & - & -1 & & & & & & + & _ c & -i & & -b & & & & & + & _ a & k & & & b & & & & + _ _ 1^ & _ b & & & & & & & & + _ ( _ 1^ ) & _ b & & & & & & & & + _ ( h_1 ) & & -h_0 & & & & & 2b & & + _ ( h_1 ) & & -h_0 & & & & 2b & & & -1 ) , [ matrixextpivotordn ] and @xmath544 in which the zeros were written only in the empty rows , for clarity , and we used that @xmath545 implies @xmath546 . + in the matrix @xmath258 , grouping together the eighth and ninth rows and columns produces a @xmath547 lower triangular matrix , whose coefficients on the diagonal are invertible , since the operator @xmath218 is invertible on gyro - fluctuations .
the eighth and ninth rows and columns have been grouped together because they constitute an invertible @xmath276 matrix on the diagonal . a convenient way to invert
the resulting @xmath547 matrix is to separate its diagonal terms : @xmath548 where @xmath549 has null diagonal and @xmath550 is purely diagonal .
then the equation can be written @xmath551 the solution of equation ( [ inductionordn ] ) is then ( _ n)_^t = ( -d ) , [ inductionordnsol ] where @xmath552 is a diagonal matrix with coefficients @xmath553 in which @xmath554 is the inverse matrix for the coupled system ( the eighth diagonal term of the 9 * 9 matrix mentioned above ) @xmath555 the operator @xmath556 is the gyro - integral operator . it can be computed without introducing any gyro - gauge , with the intrinsic calculus introduced in @xcite , or with the matrix calculus introduced in @xcite .
also , the coordinate @xmath36 can be used as an intermediate quantity for this computation , which is made at constant @xmath3 , so that the presence of a gauge ( only for the intermediate computation ) is of no consequence ; then @xmath556 is the primitive with respect to @xmath129 such that its gyro - average is zero . over the fourier modes @xmath557 ( i.e. over gyro - fluctuations )
, it is the operator @xmath558 .
+ even if @xmath559 appears in its right - hand side , formula ( [ inductionordnsol ] ) is an explicit solution for the induction relation : in the matrix @xmath560 , all the coefficients on the diagonal or above it are zero , so that when computing the unknowns one after the other starting from the left , each of them is computed as a function of previously computed quantities , i.e. the right - hand side contains only known quantities or parameters , but none of the remaining unknowns .
alternatively , the induction can be solved using a standard matrix inverse ( _ n)_^t = - _ , [ inductionordnsol2 ] with @xmath561 easily computed from ( [ matrixextpivotordn ] ) , but the coefficients are more complicated , and practical computations usually follow the procedure of formula ( [ inductionordnsol ] ) . for the solution @xmath484 , formula ( [ inductionordnsol ] ) or ( [ inductionordnsol2 ] ) must be completed by the following relations for the trivial components of @xmath94 : @xmath562 and by the determination of the parameters @xmath541 . + the first parameter @xmath540 will be determined at the following order , in an equation that does not involve @xmath509 and @xmath523 , so that there are not coupled equations between the orders .
the second parameter @xmath519 is either put to zero or adjusted so as to make the reduced hamiltonian @xmath523 zero by the relation @xmath563 \ , .
\notag\end{aligned}\ ] ] this last choice is possible only if the term inside the parentheses has no overall contribution of order zero in @xmath293 .
the last parameter @xmath443 is determined by equation ( [ seculareqordn ] ) , in order to make both @xmath519 and @xmath523 zero , when the equation is integrable .
otherwise , it is determined to cancel as many terms of @xmath519 and @xmath523 as possible , or it can be set to zero for simplicity . accordingly , at the end of each order @xmath479 , the situation is the same as at the end of order @xmath34 , with the presence of one parameter @xmath540 , of one binary choice between @xmath519 and @xmath523 , and of one free variable @xmath443 . when the integrability condition can be satisfied , the reduction of the lagrangian @xmath94 is complete and the solution is defined to within an element in the kernel of the operator ( [ seculareqordn ] )
. otherwise , @xmath540 is generally set to zero , and after the binary choice the transformation is unique , but on the whole there are two maximal reductions .
as announced in the appendix , the unicity of the transformation is determined by the integrability condition for @xmath108 , and possibly by an additional criterion for simplicity .
computations of the previous section for the guiding - center transformation and reduced hamiltonian can be summarized by formulae ( [ larmorradiusosc ] ) , ( [ resultsorderga11])-([resultsorderga12 ] ) and ( [ avgg1mu])-([resultsdeparametrized2 ] ) . as for the reduced symplectic lagrangian , it is exactly known , and is given by formulae ( [ lagrredordzerosol ] ) and ( [ lagrredord1sol ] ) , together with the prescription that all other terms @xmath564 are zero , except @xmath565 ( as well as @xmath324 ) , which can be freely chosen , for instance it can be chosen zero , or such that it makes the reduced hamiltonian @xmath523 zero by formula ( [ choicehn=0 ] ) .
these results agree with the standard results of the literature , provided the connection vector is defined as @xmath566 , which corresponds to the traditional gauge - dependent framework .
for instance , in the paper @xcite littlejohn made the choice @xmath567 , and accounting for this choice , our formulae agree with his ones . in the paper @xcite ,
the choice is @xmath568 , and again , accounting for this choice through formula ( [ lagrredord1alain ] ) for @xmath324 , our results agree with the ones of @xcite .
thus , the procedure of the previous section succeeds in obtaining the standard guiding - center reduction without introducing any gyro - gauge and using purely intrinsic coordinates .
+ the explicit induction relation ( [ inductionordnsol ] ) shows that the reduction is possible to arbitrary orders , but it also gives an explicit formula to practically compute the transformation generator order by order . at any order in the larmor radius , all that remains to do is to develop the lie derivatives involved in the term @xmath569 .
only the number of terms generated by the leibniz rule makes the process difficult to compute by hand at higher order , but the explicit induction involves few basic operations ( just exterior derivatives and matrix products ) and can easily be implemented to higher orders on a computer .
actually , as the series is a polynomial , the derivation does not rely on formal calculus but just on symbolic calculus , which is still easier to implement .
+ once the generators @xmath88 are obtained , the guiding - center coordinate transformation is given by @xmath570 the reduced lagrangian @xmath0-form is @xmath571 \sfb\cdott d\bfq - \sum_n\overline h_n dt \label{lagrredtotalsolution } \\ & = ( \bfma+\overline\ga^\sfb \sfb)\cdott d\bfq - \mu\sfa\cdott d\sfc - \overline h dt \
, , \notag \end{aligned}\ ] ] where the hamiltonian terms @xmath523 are provided by ( [ inductionordnsol ] ) , whereas the parallel lagrangian terms @xmath324 and @xmath519 are chosen freely at each order @xmath81 , and can be chosen zero .
the reduced dynamics is obtained the usual way , by computing the lagrange matrix @xmath572 , then inverting it to get the poisson matrix @xmath573 , and last computing hamilton s equations @xmath574 .
alternatively , the reduced equations of motion can be obtained by lie transforming directly the velocity vector field @xmath575 here , performing these computations is useless , since the reduced lagrangian ( [ lagrredtotalsolution ] ) completely agrees with previous results in the literature in the gauge - dependent case , and so will the reduced motion .
+ we only indicate the guiding - center poisson bracket , in which the effects of the gauge independence and of the higher - order corrections are interesting .
it is computed from the lagrange @xmath34-form , easily obtained from eq .
( [ lagrredtotalsolution ] ) in a matrix form : @xmath576 with @xmath577 where we used the fact that the order in @xmath168 indicates the expansion order , as mentioned about eq .
( [ scaling ] ) .
then the poisson bracket in matrix form writes : @xmath578 with @xmath579 where the reduced magnetic field @xmath580 is defined as usual by formula ( [ bstar ] ) .
the coefficients @xmath581 and @xmath582 are invertible since eq .
( [ coefflagrmatrred ] ) shows that they are small corrections from @xmath330 and @xmath325 , which are invertible . as a result , between two arbitrary functions of the reduced phase space @xmath583 and @xmath584 ,
the poisson bracket is @xmath585 where the symbol @xmath586 is a shorthand for the operator @xmath587 this operator does not depend of the free function @xmath58 , since for any choice of @xmath58 the definition ( [ nablastar ] ) gives the same result , which is equal to the covariant derivative @xmath28 when the gauge vector is chosen zero , i.e. @xmath588 .
the poisson bracket ( [ poissbrackred ] ) agrees with the literature , e.g. @xcite . especially , the traditional ordering is patent : the last two terms are of order @xmath589 , and correspond to the fast gyro - angle dynamics ; the second term is of order @xmath590 , and corresponds to the intermediate motion along the magnetic field lines ; as for the first term , it is of order @xmath591 , and corresponds to the slow drifts across the magnetic field lines .
remind @xmath592 , as defined in ( [ definitepsilon ] ) , is the usual small parameter of guiding - center theory .
in previous works , the reduced poisson bracket often had only three coefficients ( the ones for @xmath580 , namely @xmath593 , @xmath581 , and @xmath594 ) , either because of incidental lowest - order simplifications , or because of the choice they had performed for @xmath595 . for a general result about maximal reductions at arbitrary order ,
the poisson bracket is given by eq .
( [ poissbrackred ] ) , where the higher - order correction @xmath596 to the lagrangian impacts five coefficients @xmath593 , @xmath581 , @xmath594 , @xmath582 , and @xmath597 ( and hence @xmath580 ) , through the definitions ( [ coefflagrmatrred ] ) , i.e. as an effect of the five derivative operators @xmath598 , @xmath599 , @xmath600 , @xmath601 , @xmath602 , acting on the seminal term @xmath596 .
+ the only difference compared to previous results , besides the coordinate system and the term @xmath603 being let free , is that the gauge vector @xmath604 is replaced by the general connection @xmath605 .
this difference could impact the reduced magnetic field @xmath580 , given by _ * : = -a- b- c = ( + ^ ) + v , [ bstar ] where the vector @xmath606 is generated by the term @xmath607 in the lagrangian @xmath107 , which implies for the lagrange @xmath34-form the presence of the term @xmath608 the first equality comes by inserting the identity @xmath609 beside the wedge symbol .
the second and third equalities come because @xmath21 is orthonormal .
the fact that the vector @xmath47 depends only on @xmath3 implies the fourth equality , which can be rewritten to get the final answer .
( [ curlgaugevectintrin ] ) shows why the vector @xmath606 ( and hence also the reduced poisson bracket ) is indeed independent of the free function @xmath58 . in the reduced lagrange @xmath34-form ,
the concerned term ( [ curlgaugevectintrin ] ) is the counterpart in the gauge - independent approach of the term @xmath610 involving the gauge vector in the traditional approach .
actually , the curl of the gauge vector @xmath611 is also gauge - independent , and the corresponding term is explicitly given by @xmath612 where again , the second equality comes by inserting the identity @xmath613 beside the wedge symbol and by using the fact that @xmath40 is orthonormal .
so , in the reduced magnetic field @xmath580 , the term with @xmath58 exactly fits with the corresponding term with @xmath53 of previous results .
this is a good illustration of how the approach using the coordinate @xmath2 completely agrees with the gauge - independent part of the usual results , whereas it proceeds in a different way and never introduces the gauge @xmath41 .
indeed , in the usual approach , @xmath52 is assumed to depend only on the position , which implies the formula above . on the contrary , in the gauge - independent approach ,
the velocity is present in the first lines of ( [ curlgaugevectintrin ] ) , but it comes out from the computation that the result is naturally just a purely spatial term in the lagrange matrix . by the way , the argument above shows that @xmath606 is indeed the curl of @xmath614 , but it is not surprising since the spatial part of a term @xmath615 is given by the cross product with the curl of @xmath616 : @xmath617 thus , formula ( [ bstar ] ) can be rewritten @xmath618 \,,\ ] ] which is manifestly divergenceless . + here , for clarity and following previous works , we considered no electric field , in order to focus on the reduction mechanism .
the extension for a non - zero electric field is straightforward , because the lagrangian is changed only by the addition of [ lagrelecfield]@xmath619 , with @xmath620 the electric potential @xcite .
it affects the spatial part of @xmath621 , which plays no pivotal role in the derivation : it always appears in the right - hand side of the equations , i.e. in terms that are already known .
as a consequence , the presence of this term does not change the procedure at all . in the results ,
each term of order @xmath81 becomes a polynomial in @xmath98 ( or rather @xmath622 ) , which introduces a second parameter @xmath623 in the theory , which corresponds to @xmath624 or to @xmath625 .
a more detailed study shows that the momentum in denominators is only the perpendicular momentum @xmath626 , and that at the lowest orders the perpendicular electric field can be one order higher than the parallel electric field @xcite .
the series remains perturbative provided the associated parameter is small @xmath627 , as usual in guiding - center reductions . in the results above
, the usual expansions are recovered , but the choice of the coordinates @xmath2 and @xmath293 makes all quantities polynomial in the used coordinates and monomial in @xmath628 and @xmath168 .
this is useful to simplify the derivation , which can be considered as a symbolic - calculus algorithm based on just two operations acting on words ( polynomials ) composed from a very restricted alphabet .
such structures can also be useful when going beyond the formal expansions and considering them as asymptotic series .
for instance , the polynomiality in the cotangent of the pitch - angle is important to control the loss of accuracy of the expansion in the domains where the direction of the particle momentum is close to the direction of the magnetic field .
previous results were obviously polynomial in the variable @xmath2 , but they did not use it as a coordinate , since they replaced it by the variable @xmath129 .
in addition , the expansion in the usual small parameter @xmath629 was also present , but it did not correspond to an expansion in @xmath132 nor in @xmath19 , but only in @xmath28 , which is not a scalar quantity ; it is why keeping the quantity @xmath32 was useful to indicate the expansion order . here , the order is directly indicated by the power in @xmath159 or @xmath168 , since each order is a monomial in those quantities .
the monomiality in @xmath628 is especially interesting , because the orders in the operator @xmath28 and in the function @xmath19 have only a dimensional meaning : the term of order @xmath85 will involve terms like @xmath630 , but also terms like @xmath631 , or @xmath632 . in a similar way
, the order in @xmath19 is given by the order of the prefactor @xmath633 , but the variable @xmath19 appears in other places when it is acted upon by gradients ; then , it is compensated by a @xmath19 in the denominator , which means that gradients act only on the variable @xmath634 .
+ the use of the variable @xmath293 instead of the usual @xmath635 is crucial for the results at each order to be both polynomial in the coordinates and monomial in @xmath159 and @xmath168 .
it is a generalization of @xcite , which considered only the minimal guiding - center reduction .
it seems it had not been noticed in previous works on the full guiding - center reduction @xcite . when using the standard variable @xmath635 instead of @xmath293 , the monomiality in @xmath159 and @xmath168 is not verified : for instance , in @xmath333 , the term of order @xmath636 writes @xmath637 , which is not a polynomial in @xmath628 and which is not of the same order in @xmath159 nor in @xmath168 as the term of order @xmath638 , since this last writes @xmath639 , as is confirmed in @xcite , for instance . on the contrary , when using the variable @xmath293 , the polynomiality is verified , because of the structure of the lagrangian , of the action of derivatives , especially ( [ connectboth ] ) , and of the coefficients to be inverted for the matrix inversions involved in the derivation .
notice that the induction procedure also guarantees that all formulae will be polynomial in the variable @xmath47 .
as for the monomiality , it is easily understood from a dimensional analysis : only three independent dimensional quantities are involved in the results , e.g. @xmath28 , @xmath640 and @xmath641 .
when using the momentum coordinates @xmath642 or @xmath643 , two of them are dimensionless , and only one of them can generate the dimension of @xmath14 , namely @xmath14 or @xmath159 . for an expansion in @xmath644 , the variable @xmath132 can be involved only in the pre - factor of each order , otherwise , it could not be compensated to generate a dimensionless quantity . on the other hand ,
when using the momentum coordinates @xmath645 or @xmath646 , there is a redundancy in dimension between @xmath635 and @xmath14 ( or @xmath647 ) , which means that ratios of them are expected in order to get dimensionless quantities . in an expansion in @xmath30 , if the term of order @xmath648 is a polynomial in @xmath635 , it has to be actually a polynomial in @xmath649 or @xmath650 . as a consequence
, formulae will be sums of terms @xmath651 , which is not a monomial in @xmath14 ( or equivalently in @xmath159 and @xmath168 ) ; by the way , it is not a polynomial either , because of the terms where @xmath652 .
last , the polynomiality in @xmath293 means that each term writes @xmath653 , which agrees with the idea that the entities @xmath654 and @xmath655 come from expansions of the momentum @xmath5 , or rather the corresponding dimensionless vector @xmath656 .
in addition to the averaging reduction and the inclusion of the magnetic moment among the reduced coordinates , the goal was to obtain a reduced dynamics as strongly reduced as possible .
so , a complete achievement is obtained when all the components of @xmath657 are zero , at least for higher orders . for the procedure , it means using the seven unknowns to solve the seven corresponding equations , or rather twice these numbers , if one considers the average and the fluctuating parts as different variables .
it did work for all the requirements but one , which could not be satisfied and had to be dropped .
the obstruction for a complete reduction @xmath658 comes from the requirement ( [ requirprior2 ] ) : it imposes to obtain @xmath529 by fixing the freedom @xmath659 , which was the only freedom available for the equation @xmath660 and is no more available for it .
it is why one of the requirements has to be dropped .
then , it remains more unknowns than requirements .
so , the transformation is not unique .
especially , while it naturally appeared in @xmath135 , the non - zero component of @xmath107 can be transferred to @xmath661 .
thus two maximal reductions can be considered .
+ the first alternative ( called the _
hamiltonian _ representation in @xcite ) sets @xmath662 .
then , the reduced poisson bracket is completely known before computing the transformation to higher orders ; it is given by the lowest three orders of the symplectic lagrangian @xmath663 .
the reduced hamiltonian is not exactly known ; it is given by a whole series in @xmath30 and must be computed order by order .
the reduced dynamics is a hamiltonian perturbation of the guiding - center equations of motion at order @xmath34 . the second alternative ( called the _
symplectic _ representation in @xcite ) is to set @xmath664 .
then , the hamiltonian is completely known , and the structure of the poisson bracket is also known . the only unknown information on the reduced dynamics is concentrated in the component of the reduced lagrangian parallel to the magnetic field @xmath519 , which is given by a whole series in @xmath30 .
the poisson bracket includes a kind of reduced magnetic field @xmath580 , induced by the higher - order terms of the lagrangian .
the choice of symplectic or hamiltonian representation can be made at each order in the derivation , but it seems more convenient to be consistent and to make the same choice for all orders , as suggested in @xcite .
+ these two maximal reductions give a unified view of various choices that can be found in the literature , and they anticipate what will happen at higher orders .
indeed , even in the standard non - canonical hamiltonian approach of the guiding - center reduction introduced by littlejohn , several transformations can be found , often related by differences of choice related to this two - fold maximal reduction .
for instance , littlejohn s initial guiding - center reduction @xcite corresponded to the second possibility above at order @xmath259 , but at higher order , it is unclear whether the procedure provided a maximal reduction , or if some terms could remain both in the hamiltonian and in the poisson bracket .
the seminal reduction by lie transforming the lagrangian @xcite corresponded to a maximal reduction with the first choice ( `` hamiltonian representation '' ) .
later papers by lie transforming the lagrangian turned to the other choice , e.g. @xcite used a maximal reduction ( at order @xmath259 ) but with the second possibility . recently , while improvements in the second order were addressed @xcite , interest was renewed in the first possibility .
the paper @xcite actually corresponds to a mixed choice , where the second possibility is used at order @xmath259 and the first one is used at order @xmath369 .
the work @xcite introduced the designation of _ hamiltonian _ and _ symplectic _ representation to differentiate between the two choices , and studied an `` equivalence relation '' between the two choices ( when the same choice is made for all orders ) , which is a way to go from one representation to the other by a redefinition of the reduced parallel momentum @xmath665 ( or equivalently of @xmath666 ) . + the algorithm at higher order , with the condition ( [ choicehn=0 ] ) for the symplectic representation , had not been studied in detail . in the previous section ,
the hamiltonian representation appeared as indeed guaranteed at arbitrary order , and as naturally provided by the procedure as a maximal guiding - center reduction , whereas the symplectic representation appeared as submitted to a condition at each order in @xmath30 : in formula ( [ choicehn=0 ] ) , the term inside the bracket must have no overall contribution of order @xmath147 in @xmath293 .
this could explain why first papers addressing both the first and the second order in @xmath30 systematically used the hamiltonian representation ; the symplectic representation was used only later , when the condition was observed as verified .
the equivalent relation introduced in @xcite relies on a relationship between these representations , which allows to go from one to the other .
hence it might seem that it guarantees the existence of the symplectic representation , but it is not the case . the underlying idea ( see equation ( 17 ) in @xcite ) is the following .
start from the reduced lagrangian written in symplectic representation ( we use here the variable @xmath667 instead of @xmath293 , in order to agree with the notation used in @xcite ) : @xmath668 then , redefine the reduced parallel momentum so as to absorb all the series @xmath595 in it : p_ : = p _ + _ n=1 ( _ ) _ n .
[ transformsymplecttohamil ] with this coordinate , the symplectic part of the lagrangian is fully reduced : @xmath669 to obtain the reduced hamiltonian with this coordinate , one just inverts the series ( [ transformsymplecttohamil ] ) : p _ : = p_ - _ n1 ( _ ) _ n = p_ + _ n1 ( _ )_n , [ seriesinverseregular ] with @xmath670 some coefficients easily obtained by inserting iteratively the first equality in the occurrences of @xmath665 in @xmath671 , as is standard to invert a near - identity series .
then the reduced hamiltonian in the new coordinate writes @xmath672 ^ 2 \,.\ ] ] it is a full series in @xmath30 , which corresponds to the hamiltonian representation .
this is a constructive procedure showing that when the symplectic representation exists , then the hamiltonian representation exists and is easily obtained .
+ now , what is actually needed is to go in the reverse direction , since the derivation of the guiding - center reduction shows that the hamiltonian representation is natural and guaranteed to exist , whereas the symplectic is suspected of having existence conditions .
it turns out that the procedure in the reverse direction can break down .
start from the reduced lagrangian written in hamiltonian representation : @xmath673 then , redefine the reduced parallel momentum so as to absorb all the higher - order terms @xmath674 in the term with @xmath665 : p_^2 : = p_^2 + 2 _
[ transformhamilttosymplect ] with this coordinate , the hamiltonian part of the lagrangian is fully reduced : @xmath675 to obtain the reduced symplectic lagrangian with this coordinate , one just inverts the near - identity transformation ( [ transformhamilttosymplect ] ) by writing it as : p_ : = = p _ [ seriesinversesingular ] , and then by expanding the term @xmath676 .
this assumes that the ratio @xmath677 is small .
the point is that this condition is not guaranteed a priori , even if the series is near - identity . indeed , the derivations here are formal .
`` near - identity '' has only a dimensional meaning .
it means that the ratio between the first and zeroth - order term is of order @xmath678 , but only in dimension , its value might not to be small if it is multiplied by a large dimensionless factor such as @xmath679 , as in ( [ seriesinversesingular ] ) .
thus , a division by @xmath665 can cause a singularity .
when going from the symplectic to the hamiltonian representation , no such a division was needed , since the series inversion ( [ seriesinverseregular ] ) just consisted in composing series . on the contrary ,
when starting from the hamiltonian representation , the series inversion ( [ seriesinversesingular ] ) involves a division by @xmath680 .
this causes a singularity if @xmath523 contains a term of order @xmath147 or @xmath0 in @xmath665 .
it is interesting to see that difficulties arise here at @xmath681 ( or equally at @xmath452 ) , which is precisely where they appeared in the guiding - center reduction in the previous section .
this suggests that singularities in @xmath452 are indeed a difficulty for the symplectic representation .
accordingly , at each order , it can be used only when the absence of singularity in ( [ choicehn=0 ] ) is verified .
the intrinsic formulation of the guiding - center reduction was motivated by questions about the traditional gyro - angle variable @xmath129 .
the derivation with the coordinate @xmath2 shows that it does succeed in shedding light on those questions .
first , the traditional coordinate was a detour . in all guiding - center works
, the variable @xmath129 never appears in itself ( except in its own definition and subsequent relations ) ; for instance , it does not appear explicitly in one of the components of @xmath91 or @xmath107 , which all depend on @xmath129 only through the corresponding physical quantity @xmath2 ( or @xmath20 ) ; even the gyro - angle component of the generator @xmath682 verifies this statement .
the detour is not given by the physics , since it imposes to fix arbitrarily a gauge @xmath37 , which is not related to the physics of the problem .
the role of the intrinsic approach was to avoid this detour , and it achieves its goal since it obtains the full guiding - center results without introducing any gauge and by working purely with @xmath2 . from a mathematical point of view also ,
the use of the variable @xmath129 was not completely satisfactory , because the gyro - angle corresponds to a circle bundle @xcite .
the traditional coordinate @xmath129 makes this structure somehow disappear , because the manifold trivially becomes @xmath683 .
it is why the variable @xmath129 does not have a global existence in a general magnetic geometry @xcite . on the contrary , when using the physical variable @xmath2 , the circle bundle naturally arises : as @xmath2 is defined on a space - dependent circle , spatial displacements imply a variation of @xmath2 , so that a covariant derivative is involved , which encodes the circle - bundle geometry for the gyro - angle @xcite and does not imply some restricted class of circle bundle .
a more detailed study of the coordinate system is outside the scope of the present paper , and will be reported in @xcite . + in some way , the relevance of this coordinate is obvious , since it just results from keeping the initial coordinate , in which all the circle - bundle picture was included . from this point of view , performing the derivation with this variable is a way to see how it globally agrees with the physics and the mathematics of the problem , and to make intrinsic definitions arise naturally for all the quantities involved in the process .
indeed , the previous section shows that the reduction follows the same procedure with the vectorial constrained coordinate @xmath2 as with the scalar coordinate @xmath129 , but that there are slight changes in the quantities used .
the gauge vector @xmath53 came as naturally replaced by the connection @xmath58 for the covariant derivative .
the generator of larmor gyrations @xmath219 came with an intrinsic definition @xmath218 .
the basic @xmath0-form for the gyro - angle @xmath1 appeared as replaced by a non - closed @xmath0-form @xmath194 , which agrees with the fact that @xmath129 considers the circle bundle trivial , whereas it should not .
this implied to use more intrinsic definitions for the operations used , such as ( [ extderivgene ] ) and ( [ extderivpractical ] ) for exterior derivatives . also , this implied to be careful on how the basis of @xmath0-forms and of vector fields are chosen , but the natural ones were found to agree with each other .
thus , the formalism with @xmath2 is slightly more involved , but it perfectly fits both with the physics and the mathematics of the problem , which correspond to a non - trivial circle bundle . + from a formal point of view , the results with the physical variable @xmath2 mainly correspond to replacing the gauge vector @xmath53 by the connection term @xmath58 .
thus , they include the standard gauge - dependent results as a special case , but they are more general : in the usual approach @xmath684 depends only on the position and can not be chosen freely ( e.g. @xmath685 is not possible @xcite ) ; here , @xmath54 can be any function of the position and momentum .
especially , the physical definition of @xmath2 corresponds to the function @xmath686 , which depends also on the momentum and preserves the polynomiality in @xmath293 .
+ other gauge - dependent quantities are interesting to consider . in previous works ,
the coordinate transformation @xmath687 was gauge dependent ( see for instance equation ( 30c ) in @xcite , or in @xcite the solution for the generator @xmath688 below equation ( 5.45 ) ) , as well as the definition of the coordinate @xmath129 , and also the gradient @xmath689 .
it is why the gauge vector @xmath53 was involved in some of the resulting formulae , e.g. the poisson bracket , in such a way as to make all the physical or geometrical ( intrinsic ) quantities gauge - independent .
for instance , in the poisson bracket , gradients appear only in the combination @xmath690 @xcite .
it would be interesting to interpret it as the gradient corresponding to a special gauge , because it would remove the appearance of the gauge vector in all the derivation , and would simplify computations .
the issue is that it is not possible because it would correspond to fix the gauge in such a way that @xmath691 , which is not possible even locally @xcite . in the gauge - independent approach ,
all the coordinates , including the gyro - angle @xmath2 , are gauge - independent , as well as the transformation @xmath692 : at first order , it is not transformed by @xmath693 , because the covariant derivative must be taken into account , which means that it is given by @xmath694 .
this last quantity is indeed independent of the connection vector @xmath695 , as can be verified in ( [ solg1th ] ) . in the same way , at higher order ,
all the transformed coordinates @xmath696 will be independent of @xmath58 , where @xmath697 are physical coordinates .
gradients are also involved in combinations involving @xmath58 ( see eq .
( [ poissbrackred ] ) , for instance ) .
this is no surprise , since the connection on the fibre bundle involves some arbitrariness , but the combinations can always be written @xmath698 , which is connection - independent .
in addition , it can be interpreted as the covariant derivative associated with the trivial connection @xmath699 .
thus , when working with the coordinate @xmath2 , this choice can be used to simplify computations and to make them connection - independent .
the derivation procedure confirms the respective interests of lie transforming the velocity vector field and the lagrangian @xmath0-form . as with concerns the minimal requirements for the guiding - center transformation , working on the equation of motion is much more efficient , since it systematically obtains the fluctuating part of the reduced motion just by inverting the operator @xmath219 .
the procedure with the lagrangian is much more involved , as can be seen in previous sections , especially because the order mixing makes the scheme more elaborated and because the algorithmic stage begins only at higher order : the induction matrix mixes up the orders , changes at each order for @xmath700 , and involves differential operators in some coefficients .
it is why in this paper , as in previous works , only a part of @xmath253 is explicitly computed , whereas the work @xcite directly obtained the full second - order generator @xmath253 .
in addition , the minimal guiding - center reduction can hardly be obtained by working on the lagrangian , because going from the lagrangian to the motion mixes the components up . to guarantee an averaged slow reduced motion for the four coordinates @xmath701 ,
one would need to average all of the seven components of the lagrangian , which is not a minimal transformation . + as with concerns the additional requirements for the slow dynamics , working on the equation of motion is not efficient , because the equations to be solved are secular differential equations that are not simple to deal with @xcite .
working on the lagrangian is more efficient , because it essentially consists in algebraic equations , which deals the same way with gyro - averages as with gyro - fluctuations .
this makes it easy to identify good choices for the averaged transformation generator @xmath702 in order to obtain a reduced lagrangian as strongly reduced as possible .
thus , it provides a maximal guiding - center reduction almost as simply as the minimal one .
also , working on the lagrangian @xmath0-form makes it easy to impose requirements on the reduced hamiltonian structure , for instance to obtain a quarter - canonical structure for the coordinates @xmath141 , which both provides a constant of motion @xmath127 and a hamiltonian sub - dynamics for the @xmath139-dimensional reduced motion @xmath703 .
the full guiding - center reduction can be performed to arbitrary order in the larmor - radius expansion by lie transforming the lagrangian @xmath0-form while keeping physical gyro - gauge - independent variables as coordinates , following the same procedure as when working with the standard gauge - dependent gyro - angle . for higher orders ,
the procedure was shown to be completely algorithmic .
the pivotal role is played by the inverse of the lowest - order lagrange matrix @xmath359 , together with a differential equation for the function @xmath108 . an extended matrix was defined and used to explicitly solve the induction equation to arbitrary order in the larmor radius .
+ the results exactly agree with previous works , but they were obtained without introducing any gyro - gauge , and working purely with the physical coordinate @xmath2 as the gyro - angle coordinate .
in addition , the choice of the cotangent of the pitch - angle as a coordinate for the parallel velocity made the results purely polynomial in the coordinates and monomial in @xmath628 and @xmath168 .
+ compared to the method by lie transforming the equations of motion , the process is much more elaborated , especially because of the order mixing , but it easily obtains a much stronger result .
it does not rely on differential equations for the reduced motion , but on algebraic equations for the reduced lagrangian .
a quarter - canonical reduced hamiltonian structure provides a constant of motion @xmath127 and a hamiltonian sub - dynamics for the @xmath139-dimensional slow reduced motion @xmath704 . in addition , the procedure makes the reduced dynamics trivial not only in the gyro - fluctuating components of the lagrangian , but also in six of the averaged components out of seven . as a result ,
all but one of the components of the reduced lagrangian @xmath0-form are put to zero for all orders higher than two .
only one of them can not be made exact , and is given by a whole series .
the two canonical choices are recovered : either to enclose the series into the hamiltonian ( hamiltonian representation ) , then the reduced poisson bracket is exact , or to enclose the series into the spatial component of the lagrangian parallel to the magnetic field ( symplectic representation ) , then the hamiltonian is exact and the uncertainty of the reduced motion is traduced by five coefficients , and especially a kind of reduced magnetic field @xmath580 .
the hamiltonian representation appeared as naturally induced by the reduction process , whereas the symplectic representation is subjected to a condition at each order , to avoid a singularity in @xmath452 , i.e. at the bounce points of particle trajectories .
these representations make the reduction maximal because for a general magnetic field , the procedure can not get a stronger reduction for which even the last component of the lagrangian would be zero .
the obstruction originates from the special role of the magnetic moment ; in the hamiltonian representation , this can be viewed because the magnetic - moment component of the transformation generator can remove only the fluctuating part of the reduced hamiltonian function , since the averaged part is imposed by the requirement of adiabatic invariance .
+ the use of gauge - independent coordinates had little effect on the reduction procedure .
all the ingredients of the standard reduction with gauge - dependent coordinates were found to be present , but they naturally arose with an intrinsic definition or they were replaced by a different intrinsic object playing a similar role .
it was observed to fit in with both the physics and the mathematics of the system , by restoring the general circle - bundle framework , which practically disappeared with the coordinate @xmath129 , and by making the coordinates directly induced by the physical state of the system .
for instance , the gauge removal introduced a vectorial quantity @xmath2 for the gyro - angle coordinate .
this caused the coordinate system to be constrained and implied a connection for the covariant derivative on a space - dependent circle , which is directly linked to the circle - bundle structure underlying in the gyro - angle coordinate and which replaced the gauge vector of the gauge - dependent approach .
the closed @xmath0-form @xmath1 was replaced by a non - closed @xmath0-form @xmath194 , which is related to the non - triviality of the circle bundle for a general magnetic geometry . in previous works relying on the coordinate @xmath129 , the gauge - independence of the physical results implied that gradients were systematically involved in special expressions , which were not related to derivative operators because no gauge fixing were suited to them .
these expressions were found to be related to covariant derivatives corresponding to suitable connections .
+ unlike the gauge fixing for the coordinate @xmath129 , the connection fixing for the coordinate @xmath2 depends not only on the position , but on the momentum as well .
this is all the more convenient as the physical definition of @xmath2 and its associated connection depend on both the position and the momentum . in addition
, this removes one of the assumptions causing the presence of anholonomy in the gyro - angle motion .
so , the intrinsic gyro - angle coordinate @xmath2 is a way to tackle some of the questions involved in the guiding - center anholonomy and gauge- ( or connection- ) arbitrariness .
these questions are outside the scope of this paper and will be reported elsewhere @xcite .
+ in this paper , we focused on the formal derivation of perturbation series , as is usual in guiding - center works , and as is the standard first step in perturbation theory @xcite
. a next step will be to relate these formal expansions with asymptotic series , in a similar way as what was done in @xcite for kruskal s work , and what is beginning being done about littlejohn s works @xcite .
also , as usual in perturbation theory , convergence of the guiding - center series is an interesting question to investigate , probably with methods of accelerated convergence @xcite . in these attempts ,
the structures of the expansion series , such as the polynomiality induced by the cotangent of the pitch - angle , are expected to play a role .
we acknowledge financial support from the agence nationale de la recherche ( anr gypsi ) .
this work was also supported by the european community under the contract of association between euratom , cea , and the french research federation for fusion study .
the views and opinions expressed herein do not necessarily reflect those of the european commission .
the authors also acknowledge fruitful discussions with alain brizard , phil morrison , mathieu lutz and with the quipe de dynamique nonlinaire of the centre de physique thorique of marseille .
in this appendix , we introduce the mechanism at work when lie transforming the lagrangian @xmath0-form . indeed , the basic ideas of the derivation are very elementary , but they are hidden by the details of the procedure , which are rather involved because of some order mixing and other subtleties between algebraic and differential integrability conditions .
in addition , practical computations in the case of the guiding - center are somehow intricate .
all the same , the method is very efficient and has quite a wider domain of application than just the guiding - center reduction .
so , it seems useful to give a general overview of the method for people not familiar with it .
\a ) the goal is to solve equations ( [ reduclagr ] ) for the unknowns @xmath88 and @xmath108 , with the requirements ( [ requirprior1 ] ) , ( [ requirprior3 ] ) and ( [ requirprior2 ] ) identified above .
ideally , the maximal reduction sets @xmath517 for all higher orders @xmath81 , as a result of ( [ requirprior3 ] ) .
the solution is built order by order in the larmor radius .
each order implies to solve the equation @xmath705 where @xmath486 is a shorthand for all other terms , that do not contain the highest - order generator @xmath488 .
the very basic idea of the reduction is that @xmath488 is involved only through a matrix product .
so , at any order , the solution is just given by a matrix inversion _
n+1 = ( _ -1)^-1 , [ basicidea ] provided @xmath228 is invertible ; then @xmath94 can be chosen zero , and @xmath108 is not useful and can also be set to zero .
this idea of matrix inverse is the key ingredient of the underlying mechanism , even if the corresponding basic picture is not true at the lowest orders , and at higher orders , it is slightly complicated by some order mixing and integrability conditions , especially for @xmath108 .
+ b ) as a matter of fact , the matrix @xmath228 is usually not invertible , since it corresponds to the fast part of the dynamics , here the larmor gyration , which does not concern all the phase - space coordinates .
+ at zeroth order , under the requirement @xmath517 , equation ( [ reduclagr ] ) writes @xmath706 it has a solution only if the right - hand side @xmath707 is in the range of the matrix @xmath228 ( solvability condition ) ; this is a necessary condition for the corresponding reduction to exist .
usually , it is not verified for so strong a requirement as @xmath708 , and the reduced lagrangian @xmath227 has to be used as a softening parameter
. then equation ( [ reduclagr ] ) writes _
1_-1 = _ 0-_0 - ds_0 .
[ equag1matrix ] one has to check that , with the freedoms @xmath226 and @xmath227 , the solvability condition can be satisfied at least for the minimal requirement ( [ requirprior1 ] ) and if possible for the intermediate requirement ( [ requirprior2 ] ) ; then , the reduction is possible , and the maximal requirement ( [ requirprior3 ] ) can be considered by trying to remain as close as possible to the condition @xmath708 . at that point ,
the solution exists , but it is not unique ; it is defined to within an element of the kernel of @xmath228 .
the choice of this element may be free at this stage of the reduction , but care must be taken that it may be constrained by the solvability conditions at the following order .
+ at the next order @xmath259 , equation ( [ reduclagr ] ) writes _ 2_-1 + ( _ 0+_0 ) = _
1- ds_1 . [ eqforx2 ]
now , the pivotal matrix @xmath709 to be inverted is the set of @xmath228 and @xmath710 , acting on the set of unknown components of @xmath263 .
its rank is greater than ( or equal to ) the rank of @xmath228 .
care must be taken that some of the coordinates of @xmath225 are already determined .
this introduces some order mixing , where some components of @xmath88 are determined at order @xmath711 , others are computed at order @xmath712 at the same time as some of the components of @xmath488 .
notice that @xmath709 can not be invertible on the unknown components of @xmath253 , because of the non - trivial kernel of @xmath228 . in the same way
as at zeroth order , this kernel has to be excluded when studying the invertibility of @xmath709 , because it will be involved only at the following order . then , if the pivotal matrix @xmath709 is invertible for the ideal requirement @xmath713 , then the solution exists and is unique
otherwise , there is again both a solvability condition and a non - uniqueness of the solution .
more precisely , the solvability condition means that there is a solution only if the right - hand side is in the range of the matrix to be inverted , and that in order to fulfil this condition , the reduced lagrangian @xmath714 may have be chosen non - zero , but having as many null components as possible .
the non - uniqueness means that the solution is determined only to within an element of the kernel of the pivotal matrix @xmath709 .
the choice of this element may be free , but care must be taken that it may be constrained by the solvability condition at the following order .
+ at the next orders , the same process goes on .
the pivotal matrix @xmath715 evolves at each order and its rank increases to determine more and more of the unknowns . at high orders
, it becomes of constant rank , and actually it becomes the same at each order .
the critical value of @xmath81 at which this occurs will be denoted by [ definitnc]@xmath366 , and the corresponding pivotal matrix will be denoted by [ definitminfty]@xmath258 .
so , for @xmath716 , the pivotal matrix verifies @xmath717 , whereas for @xmath718 , it verifies @xmath719 .
this can be explained as follows : equation ( [ reduclagralg ] ) generically ( i.e. at high orders ) writes @xmath720 \ga_{-1 } \ncr + & \left [ ( \bfmg_n\cdott d ) + ( \bfmg_{n-1}\cdott d ) ( \bfmg_1\cdott d ) + ... + \tfrac{(\bfmg_1\cdott d)^n}{n ! } \right ] \ga_{0 } + ds_n \ , . \label{redlagrequageneric}\end{aligned}\ ] ] in this analysis , low orders are excluded because there would be some additional coefficients coming from the exponential series : for instance for @xmath259 , the term @xmath721 has a factor @xmath722 and is confounded with the last term @xmath723 . denoting @xmath724 by @xmath725 , and grouping together the highest - order lie derivatives , which contain the unknowns ( which are some of the components of @xmath726 )
, the previous equation becomes @xmath727 \label{orderforinvertible } \\ & + ... \ncr & + \sfmg_1 \left [ \tfrac{(\sfmg_1)^{n}}{n+1 ! } \ga_{-1 } + \tfrac{(\sfmg_1)^{n-1}}{n ! } \ga_{0 } \right ] + ds_n \ , .
\notag\end{aligned}\ ] ] using ( [ reduclagr ] ) for the lowest - orders reduced lagrangian @xmath728 , that are already known , the previous formula can be rewritten @xmath729 + ds_n \ncr & = \bfmg_{n+1}\cdott\overline\om_{-1 } + \bfmg_{n}\cdott\overline\om_{0 } + ... + d s_n + \sfmr_n \ , , \label{matrpivot}\end{aligned}\ ] ] where @xmath486 indicates all other terms , that are already known , since they do not involve @xmath726 .
equation ( [ matrpivot ] ) shows that the pivotal matrix @xmath715 is given by the set of matrices @xmath730 , @xmath204 , etc .
, acting on the set of unknown components of @xmath726 .
the matrix @xmath715 is the same at any ( high ) order .
it is exactly given by the reduced lagrange matrix at lowest orders .
as a consequence , the pivotal matrix @xmath258 for all high orders is identified as soon as the set of @xmath731 is observed to be invertible on the set of unknown components of @xmath726 , i.e. as soon as @xmath732 becomes invertible .
we will call [ definitnb]@xmath365 the order such that @xmath362 is invertible , whereas @xmath733 is not . at that point
, the basic picture of a ) has become a simple picture b ) , which includes two stages . at low orders it consists in dealing with non invertible pivotal matrices changing at each order , and in choosing @xmath94 such that it both fulfils
the requirement and leads to an interesting invertible matrix @xmath362 . then at high orders , the induction becomes just a matrix inversion @xmath561 , as in the initial basic picture a ) .
+ c ) the simple picture b ) has to be refined . between these two stages
, an intermediate stage takes place , since usually @xmath734 . for @xmath735 ,
the higher - orders pivotal matrix is already known but not yet efficient .
the reason is that formulae ( [ redlagrequageneric])-([matrpivot ] ) hold only for orders that are high enough , because of the coefficients generated by expanding the exponentials .
if the factors involving @xmath736 have some coefficients non unity , then formula ( [ matrpivot ] ) does not hold , which spoils the conclusion .
but this concerns only low orders .
formula ( [ orderforinvertible ] ) shows that the coefficients will be unity as soon as @xmath737 , which means that @xmath738 . + as an example , consider @xmath739 for @xmath214 . when computing the next order @xmath740 , then @xmath227 is already known , but it is not yet efficient : in formula ( [ matrpivot ] ) @xmath741 the operator to be inverted is not just the set of @xmath742 because of the factor @xmath722 in the first line , which comes because the generator @xmath225 outside the parenthesis has the same order as the generator @xmath225 inside the parenthesis .
for all higher orders , this will not happen , as is illustrated by the next order @xmath743 where @xmath521 is used for `` other terms '' , in order to avoid writing uninteresting terms .
+ d ) the order mixing can also slightly complicate the picture of c ) , by spoiling the linear algebraic framework , mainly at order @xmath259 .
indeed , the first equation to be solved for @xmath253 is ( [ eqforx2 ] ) .
however , if some of the components of @xmath225 are still not determined at that point ( this is fairly general as @xmath228 is usually not invertible ) , then they can be involved in a differential equation .
indeed , @xmath227 can be undetermined at that point , and equation ( [ eqforx2 ] ) must be let under its initial form ( [ reduclagr ] ) @xmath744 which is now a differential equation for @xmath225 , and may even be non - linear in the unknown components of @xmath225 .
this can make the scheme much more complicated : even solvability conditions may be difficult to identify .
+ e ) finally , one last point has to be taken into account as well and still makes the scheme more elaborated than the picture d ) above .
the pivotal matrix @xmath715 determines the unknown components of the generator @xmath745 , but this can generate non - zero time - component @xmath746 for the generator . for a time - independent transformation , the requirement @xmath747 , reduces the dimension of the effective generator @xmath88 .
then the pivotal matrix can be inverted only if some integrability conditions are fulfilled .
another way of saying it is that @xmath94 has seven components ( seven requirements ) whereas @xmath88 has only six freedoms .
the additional freedom comes from the gauge function @xmath108 . actually , the presence of this integrability condition for the pivotal matrix @xmath258 is completely general and comes because @xmath258 is antisymmetric .
it is not invertible on the @xmath62-dimensional space @xmath422 , and can be invertible only on a sub - space , e.g. on the @xmath67-dimensional phase - space @xmath748 . for a symplectic hamiltonian system ,
the high - orders pivotal matrix @xmath258 is indeed invertible when restricted to the phase - space , since the lagrange @xmath34-form @xmath69 is invertible , and so is @xmath749 .
so , the gauge function @xmath108 is not determined by the algebraic matrix inversion , but by the solvability condition for the matrix inversion .
furthermore , it appears in a differential equation .
existence of solution for this differential equation can involve other integrability conditions .
for instance , in an equation such as @xmath750 inverting @xmath751 implies the function @xmath752 to have no gyro - average . as a result , both the algebraic and the differential integrability conditions must be played with so as to make @xmath517 .
if it is not possible , one has to choose a non - zero reduced lagrangian @xmath753 .
this means playing with the requirements also , and releasing them slightly , so as to make the integrability conditions fulfilled and at the same time to keep @xmath107 as strongly reduced as possible .
+ all these features do not spoil the algorithmic character of the reduction for high orders , because the differential scheme is very simple ( the operators are just @xmath754 ) and in addition , at all @xmath716 , the algebraic scheme for @xmath281 is fixed , which makes it possible to conclude about the differential scheme for @xmath108 so as to make the resulting reduction maximal . at the end
, the induction relations can be written in matrix form provided some coefficients of the matrix are differential operators .
by such a redefinition of the matrix @xmath258 , the induction relation for high orders @xmath716 just relies on a matrix inverse @xmath755 .
then , the basic picture of formula ( [ basicidea ] ) becomes efficient : equations ( [ reduclagr ] ) are solved at arbitrary order through a formula completely analogous to ( [ basicidea ] ) , even if the framework is much more elaborated .
we want to stress this fact because the order mixing and the presence of integral operators may hide the triviality of the induction mechanism .
the previous subsection shows that the reduction is performed in three stages .
the first stage corresponds to the first few orders @xmath756 .
the work consists in verifying that the freedoms can be used both to make the solvability conditions satisfied and to get an interesting invertible matrix @xmath258 . at the end @xmath757 ,
the invertible high - order pivotal matrix @xmath258 becomes identified , and the first stage is ended .
the second stage corresponds to a transition stage .
the pivotal matrix for high order is identified , but it is still not efficient at that order .
the goal is only to check that the solvability conditions can be satisfied at these intermediate orders .
the third stage begins at order @xmath758 , i.e. as soon as the matrix to be inverted becomes @xmath258 . from that order on ,
it is sure that the reduction can be performed to any order in the larmor radius .
as the matrix is now invertible , the solution exists and is unique to each order , and the process becomes fully algorithmic . + in order to get a formula analogous to ( [ basicidea ] ) , the pivotal matrix must be extended to include the gauge function @xmath108 , and some coefficients of the inverted matrix @xmath561 are then integral operators .
in addition , in order to deal with the order mixing , some intermediate quantities must be introduced to isolate the components of @xmath759 that are already known from the ones that are not identified yet .
+ for example , if the pivotal matrix @xmath258 involves only @xmath730 and @xmath505 .
then , the equation ( [ reduclagr ] ) or ( [ matrpivot ] ) writes _
n= _ n+1_-1 + _ n_0 + d s_n + _ n , [ equhighord ] where @xmath486 indicates all terms of ( [ reduclagr ] ) , that do not depend on the unknowns , which are the gauge function @xmath108 and some components of @xmath760 .
these last quantities can be grouped into one single vector _
n:= ( _ n+1,_n , s_n ) .
[ defvectbfgapp ] the pivotal matrix @xmath715 is then extended to act on all @xmath484 ( including the gauge function ) in ( [ equhighord ] ) and is defined by @xmath761 as announced , some of its coefficients ( the ones acting on the component @xmath108 ) are differential operators . with these conventions , equation ( [ equhighord ] )
n= _ n_n + _ n . now , some of the components of @xmath760 are already identified at that order .
let us denote them by the index @xmath183 , and the remaining components of @xmath762 , which are not identified are denoted by the index @xmath763 : @xmath764 with @xmath765 fully identified and all terms of @xmath766 fully unknown .
the lagrangian writes _
n= ( _ n)_a(_n)_a + ( _ n)_(_n ) _
+ _ n , [ decomposbfgappen ] with obvious definitions for the linear operators @xmath767 and @xmath768 . by assumption ,
the quantities @xmath486 and @xmath769 are known ; in addition , @xmath770 is known and invertible . as a consequence ,
the induction relation writes ( _ n ) _
[ solutexpl ] it is explicit and makes the basic picture ( [ basicidea ] ) apply to all orders @xmath716 .
some coefficients of @xmath561 are integral operators , since in the inverse matrix @xmath258 some coefficients are differential operators .
a few comments are in place .
first , some components of @xmath488 remain non - identified after the order @xmath81 ; they must be excluded from @xmath766 to get an invertible matrix , because they are elements of the kernel of @xmath228 and will be determined at the next order ; this is well illustrated by ( [ defvectbfgtext ] ) and ( [ defvectbfginftytext])-([defvectbfgalpha ] ) .
second , the components @xmath765 can be extracted from @xmath484 and its term @xmath769 can be grouped with @xmath486 ( see formula ( [ defsfmrprime ] ) ) , which plays the same role .
last , the reduced lagrangian @xmath94 is in principle taken to be zero , but it was kept free because integrability conditions for @xmath108 can make it necessary to choose some of its components non - zero ; then , it can be included in the vector @xmath484 , as is done in ( [ defvectbfgtext ] ) and ( [ defvectbfginftytext])-([defvectbfgalpha ] ) .
+ the final algorithm to be iterated for the @xmath81-th - order term is trivial : in formula ( [ solutexpl ] ) , replace the lowest orders terms by their expression , already known , then compute the lie derivatives involved in the term @xmath486 , and last apply the matrix product with @xmath755 .
the mechanism involves just two kinds of operations , derivatives and a matrix product , which can be easily implemented to arbitrary order using computer - assisted formal calculus .
+ the basic idea shown in ( [ basicidea ] ) and ( [ solutexpl ] ) explains why lie transforming the lagrangian @xmath0-form has the advantage of algebraic equations , which makes it easy to reduce also the averaged part of the reduced motion , and thus to get non - minimal guiding - center reductions .
indeed , computations for the non - minimal requirements are treated the same way as for the minimal ones , the only difference concerns the priority : if all requirements can not be satisfied , then the order of priority may impose the requirements to be preferred and the ones to be released .
this is an essential advantage of lie transforming the lagrangian .
but the overall process is much more involved than the method relying on a lie transform of the equations of motion .
this last has the essential advantage of relying on just a gyro - integral , which makes it much more efficient to work on the fluctuating part of the reduced dynamics and to perform the minimal guiding - center reduction , as is clear in @xcite . in both cases
, the reduction relies on explicit induction relations , but when working with the lagrangian , the algorithmic stage ( third stage introduced above ) is efficient only for higher orders . for lowest orders , the reduction is not systematic at all , the choices are crucial to make the reduction work or not , but they must be guessed rather than derived .
in addition , many solvability conditions appear in the process , and there is no a - priori guarantee that they can be satisfied . + in the case of the guiding - center reduction , good choices appear rather naturally , solvability conditions come as easily satisfied , and the reduction can be considered as rather straightforward , but two specificities must be taken into account . indeed , as expected , at each order , the fluctuating part of @xmath88 is imposed by the minimal requirement ( [ requirprior1 ] ) , which means to put to zero the gyro - fluctuating part of the lagrangian ; and the averaged part is imposed by the other requirements ( [ requirprior3 ] ) and ( [ requirprior2 ] ) , which mean to put to zero the averaged part of the lagrangian as well ( except that @xmath288 ) .
however , one of the components of @xmath281 that remains not identified is already present in equation ( [ equhighord ] ) : @xmath771 remains as a parameter in the right - hand side of ( [ solutexpl ] ) .
furthermore , the integrability conditions on @xmath108 can not be fully satisfied , one of the optional requirements ( [ requirprior3 ] ) must be dropped ; so , the average component @xmath595 ( or alternatively @xmath772 ) is not zero but used to make the integrability condition satisfied .
accordingly , one of the freedoms ( the average gauge function @xmath773 ) remains undetermined for the maximal reduction . to determine it , a prescription must be added . for the simplest maximal reduction
, it is put to zero .
all these features will suggest to define and decompose the vector @xmath484 in a different way as in ( [ defvectbfgapp ] ) and ( [ decomposbfgappen ] ) , by including in this vector only the unknowns that are involved at that order ( see formula ( [ defvectbfgtext ] ) ) , and by distinguishing between the unknowns that will be identified and the ones that will remain parameters ( see formulae ( [ defvectbfginftytext])-([defvectbfgalpha ] ) ) . | 3ex * abstract : * the guiding - center reduction is studied using gyro - gauge - independent coordinates .
the lagrangian @xmath0-form of charged particle dynamics is lie transformed without introducing a gyro - gauge , but using directly the unit vector of the component of the velocity perpendicular to the magnetic field as the coordinate corresponding to larmor gyration .
the reduction is shown to provide a maximal reduction for the lagrangian and to work to all orders in the larmor radius , following exactly the same procedure as when working with the standard gauge - dependent coordinate .
the gauge - dependence is removed from the coordinate system by using a constrained variable for the gyro - angle .
the closed @xmath0-form @xmath1 is replaced by a more general non - closed @xmath0-form , which is equal to @xmath1 in the gauge - dependent case .
the gauge vector is replaced by a more general connection in the definition of the gradient , which behaves as a covariant derivative , in perfect agreement with the circle - bundle picture .
this explains some results of previous works , whose gauge - independent expressions did not correspond to a gauge fixing but indeed correspond to a connection fixing .
in addition , some general results are obtained for the guiding - center reduction .
the expansion is polynomial in the cotangent of the pitch - angle as an effect of the structure of the lagrangian , preserved by lie derivatives .
the induction for the reduction is shown to rely on the inversion of a matrix which is the same for all orders higher than three .
it is inverted and explicit induction relations are obtained to go to arbitrary order in the perturbation expansion .
the hamiltonian and symplectic representations of the guiding - center reduction are recovered , but conditions for the symplectic representation at each order are emphasized .
_ keywords _ : guiding center , lie transform , lagrangian @xmath0-form , maximal reduction , explicit induction , gauge - independent coordinates , gyro - gauge , connection , covariant derivative , principal circle bundle . |
at low temperatures or under pressures , molecular oxygen is solidified by weak intermolecular interactions . at zero pressure
, oxygen transforms to monoclinic @xmath0-o@xmath1 through @xmath3- and @xmath4-o@xmath1 by cooling . in the @xmath0 phase ,
oxygen molecules condense with its molecular axis perpendicular to the basal plane of the monoclinic lattice . as shown in figure 1(a ) , experiments demonstrated that the unit cell includes two oxygen molecules and the structure belongs to @xmath5 space group @xcite .
the ground electronic state of the oxygen molecule is spin triplet state . the molecular spins , which are perpendicular to the molecular axis , order antiferromagnetically on the basal plane of @xmath0-o@xmath1 with easy axis parallel to the @xmath6-axis .
the basal plane is illustrated in figure 1(b ) .
arrows in the figure show the directions of magnetic moments .
the crystal structures of solid oxygen might be correlated with the magnetic moments even at high pressures .
a raman experiment reported a phase transition from @xmath0-o@xmath1 to another monoclinic or orthorhombic structures below 3 gpa @xcite .
x - ray diffraction experiments using high - brilliance synchrotron radiation source , however , reported different phase diagrams .
et al . _ reported @xmath0-o@xmath1 transforms to @xmath7-o@xmath1 directly at 7.2 gpa at 19 k @xcite .
although they also found an anomaly in lattice constants just before the transition to @xmath7-o@xmath1 , it was not considered as a sign of a transition to other phases .
et al . _ also reported the stability of @xmath0-o@xmath1 up to about 5.3 gpa , and they found orthorhombic @xmath2-o@xmath1 ( space group : _ fmmm _ ) above 5.3 gpa at 65 k @xcite .
although the monoclinic @xmath0 structure is ascertained to be stable up to about 5 gpa at low temperatures in both experiments , the stable structure at pressures between 5 - 7 gpa is considered to be still open question . for higher pressures
, recent x - ray diffraction experiments have revealed that @xmath7-o@xmath1 consists of o@xmath8 clusters @xcite .
the @xmath7 phase transforms to the metallic @xmath9 phase at higher pressure ( @xmath10 100 gpa ) and finally to a superconducting state @xcite .
the mechanism of the transition from or to the @xmath7 phase and the structure of the @xmath9 phase are still unknown . regarding theoretical studies , etters _
_ obtained the structure of @xmath0-o@xmath1 using semi - empirical pair - potentials including magnetic interactions between o@xmath1 pairs and predicted a phase transition from the @xmath0 to an orthorhombic phase at 2.3 gpa @xcite .
we performed theoretical calculations using _ ab - initio _ pair - potentials and reported that the monoclinic @xmath0 structure is stable up to 6 gpa @xcite .
a part of the results will be presented in this paper .
first - principles investigations based on density functional theory ( dft ) @xcite were performed especially for higher pressure phases @xmath7 and @xmath9 @xcite .
the predicted structures for insulating @xmath7-o@xmath1 is however not consistent with the experimental one .
this is caused by a failure of the local density approximation ( lda ) in describing the magnetic interaction between oxygen molecules @xcite . because the failure does not influence the metallic state calculations ,
the structural relaxation based on dft with an appropriate starting structure can predict the structure of the @xmath9 phase .
a recent first - principles study reported a plausible structure for the metallic @xmath9 phase @xcite . in this paper , we study the insulating @xmath0 and @xmath2 phases . the total energy is evaluated as a superposition of pair - potentials obtained by quantum chemistry calculations including configuration interactions .
under low pressures , solid oxygen consists of weakly bound oxygen molecules . therefore the interactions between molecules need to be reproduced exactly in order to calculate the structure .
the total energy calculation based on dft is often used to investigate structures of various crystals .
unfortunately , dft calculations within the lda ( or with gradient corrections ) does not give correct intermolecular interactions for magnetic oxygen molecules .
it is related with the symmetries of the highest occupied molecular orbital ( homo ) and lowest unoccupied molecular orbital ( lumo ) , and the well - known bandgap - underestimation problem of the lda @xcite .
both of the homo and lumo of interacting oxygen molecules are originated in the @xmath11 ( gerade ) orbital of o@xmath1 .
thus the underestimation of the homo - lumo gap cause a serious overestimation of the exchange energy .
we actually confirm dft calculations do not reproduce the @xmath0 structure at zero pressure . in many cases , quantum chemistry calculations including
the configuration interactions reproduce intermolecular interactions correctly . in this paper
, we evaluate the total energy as a superposition of intermolecular potentials ( pair - potentials ) calculated by an _ ab - initio _ method with configuration interactions . in order to reduce degrees of freedom of the o@xmath12 system
, we introduce following assumptions into the calculations . 1 . an antiferromagnetic ordering in the @xmath13 ( basal ) plane .
2 . the @xmath5 lattice symmetry .
3 . a molecular axis perpendicular to the @xmath13 plane .
these assumptions are consistent with known properties of the structure of @xmath0-o@xmath1 , and we can treat the orthorhombic @xmath2 structure as a special case that the monoclinic angle @xmath14 . in the orthorhombic structure ,
two types of the magnetic ordering along the @xmath15-axis are allowed : the ferromagnetic or anti - ferromagnetic orderings . in the @xmath0 phase ,
the magnetic ordering along the @xmath15-axis is obviously ferromagnetic because of the periodicity along @xmath15-direction .
although recent neutron diffraction experiment reported anti - ferromagnetic ordering along the @xmath15-axis in the @xmath2 phase @xcite , we assume the ferromagnetic ordering along the @xmath15-axis because it simplify the treatment of the transition from @xmath0 to @xmath2 phases .
we have however confirmed that the difference in the total energy between the two kinds of orderings on the same lattice is very small .
it is ascribed to the weak magnetic interaction between molecules in different @xmath15-layers . following the assumption 1 , there are two kinds of pairs of oxygen molecules .
one is the pair in which magnetic moments are parallel ( ferromagnetic pair , f ) .
the other is the pair in which magnetic moments are anti - parallel ( antiferromagnetic , af ) .
assumptions 2 and 3 restrict geometric configuration of molecules .
in addition to these assumptions , we fix the inter - atomic distance in o@xmath1 . by these constraints ,
degrees of freedom of the o@xmath12 system , in other words the number of geometric parameters of pair - potentials , are reduced from twelve to two .
remaining degrees of freedom are the relative position of molecules .
the relative position of oxygen molecules ( @xmath16 ) is defined as given in figure 2 .
consequently , the total energy is written as @xmath17 ~. \label{etot}\ ] ] in the equation , pair - potentials @xmath18 and @xmath19 denote interactions of the f and af pairs , respectively . @xmath20 and @xmath21 are indices of lattice vectors and coordinates @xmath22 and @xmath23 are given as @xmath24 here , @xmath25 and @xmath4 are lattice constants and the monoclinic angle , respectively
. pair - potentials @xmath18 and @xmath19 are calculated by a complete active space self - consistent field ( casscf ) method @xcite .
the casscf calculation is performed with gamess program @xcite using 3s2p1d atomic natural orbital basis set @xcite .
details of calculations and obtained pair - potentials have been given in our previous brief report @xcite .
a raman experiment reported @xmath0-o@xmath1 transforms to another monoclinic structure or orthorhombic structure under 3 gpa @xcite .
x - ray diffraction experiments by akahama _ et al . _ and gorelli _ et al .
_ , however , showed the @xmath0 structure is stable up to higher pressure . in this section ,
we present a result of the theoretical calculations that the @xmath0 structure is stable up to 6 gpa .
the calculated pressure dependences of lattice parameters up to 6 gpa are shown in figures 3(a),(b ) and ( c ) .
the results of higher pressure , which is also shown in the figures , will be discussed later . in figures , solid lines
present calculated lattice parameters , and closed circles , triangles and diamonds are experimental results @xcite .
the values of pressure are evaluated by the numerical differentiation of the total energy with respect to the volume ( @xmath26 ) .
as previously mentioned , raman and theoretical studies reported phase transitions in this pressure range @xcite .
the structural transition predicted by the theoretical calculation was accompanied by abrupt changes of lattice parameters at 2.3 gpa .
as shown in figure 3 , present results decrease continuously with increasing pressure and there are no sign of the structural transition to the @xmath2 structure up to 6 gpa . as shown in figure 3(a ) , the crystal is less compressible along the @xmath6-axis than the @xmath27-axis .
the relation of the anisotropic lattice compressibility and the af ordering in the @xmath13 plane is discussed by akahama _
the @xmath28 ratio is 0.638 at zero pressure , and it increases with pressure and reaches 0.691 at 6 gpa . the experimental value is 0.635 at zero pressure @xcite , and it reaches 0.687 at 6 gpa @xcite .
the agreement between the calculation and experiments is very satisfactory .
as previously reported , the crystal structure of @xmath0-o@xmath1 is considered as a result of magnetic interactions between oxygen molecules @xcite . in order to clarify how the magnetic interactions affect the crystal structure , we consider a simple model of the @xmath13 plane as shown in figure 4 .
this figure shows the top view of the @xmath13 plane .
the molecular axes are perpendicular to the figure , and arrows represent magnetic moment directions of o@xmath1 molecules .
the unit cell contains ten pair of oxygen molecules : four af pairs at a distance @xmath29 and three kinds of f pairs at distances @xmath27 , @xmath6 and @xmath30 . for simplicity , we neglect interactions between molecules in which intermolecular distances are @xmath30 . consequently the ( total ) energy of this system is written as @xmath31 ~,\end{aligned}\ ] ] where @xmath32 and @xmath33 denote interactions ( pair - potentials ) of the af and f pairs at a distance @xmath34 . calculated intermolecular potentials @xmath32 and @xmath33 , which correspond to @xmath35 and @xmath36 in ( [ etot ] ) , are shown in figure 5 .
optimal energy of ( [ etotmodel ] ) are obtained at ( @xmath37 ) = ( 5.42 , 3.41 , 3.20 ) , and those are indicated by arrows in the figure .
the obtained values are very close to the experimental values at zero pressure ( 5.40 , 3.43 , 3.20 ) @xcite .
evidently ( [ etotmodel ] ) is dominated by the first term because of its coefficient and the fact that @xmath32 has deeper minimum than @xmath33 .
thus we obtain the optimal value of @xmath38 , @xmath39 where @xmath40 takes the lowest energy .
then the problem is simplified as @xmath41 , where @xmath42 and @xmath43 .
@xmath44 takes a minimum on the condition @xmath45 , i.e. , @xmath46 this condition is symmetric for @xmath6 and @xmath27 , then @xmath47 satisfies the condition .
it gives , however , the highest energy because @xmath48 is negative at @xmath49 as shown in figure 5 .
another solution @xmath50 of ( [ cond ] ) is located at slightly larger side of the minimum of @xmath33 : around 3.4 . as a result ,
the optimal value of @xmath38 and @xmath6 , the structure of the @xmath13 plane therefore , strongly depend on the position of the minimum of @xmath32 and @xmath33 .
the difference of @xmath32 and @xmath33 corresponds to the exchange energy . if there are no magnetic interactions between the molecules , @xmath32 is equivalent to @xmath33 . in this case , following the above discussion , the optimal value of @xmath38 is equal to @xmath50 and it gives the close - packed triangular lattice .
the distortion is not introduced if there are only one kind of pair - potential .
namely the origin of the distorted triangular lattice in the @xmath0 phase is attributed to the existence of the two kinds of pair - potentials , in other words , the magnetic interaction .
the anisotropic lattice compressibility of the @xmath13 plane , which is pointed out by akahama _
et al . _
@xcite , can be easily understood from figure 5 .
first we consider a situation that the lattice is compressed along the @xmath27-axis and the length along the @xmath6-axis is fixed . in this case , @xmath38 decrease following the relation @xmath51 and the first term of ( [ etotmodel ] ) increases by the repulsive interaction .
some part of the increased energy is however cancelled out by the term of @xmath52 because it is in the attractive region . on the other hand , when the lattice is compressed along the @xmath6-axis and @xmath27-direction is fixed , both term of @xmath40 and @xmath53 increase , and as a result the total energy @xmath54 increase steeply compared with the previous case .
this is the origin of the anisotropic lattice compressibility of the @xmath0 phase .
x - ray diffraction experiments by akahama _
et al . _ and gorelli _ et al .
_ showed the monoclinic @xmath0 structure is stable up to higher pressure than the previously proposed transition pressure . around 6 gpa , however , they proposed different structures .
_ reported a direct structural transformation from @xmath0- to @xmath7-o@xmath1 at 7.2 gpa . on the contrary ,
gorelli _ et al .
_ found @xmath2-o@xmath1 between @xmath0- and @xmath7-o@xmath1 . in this section
, we discuss the stability of @xmath0- and @xmath2-o@xmath1 . in figure 3 ,
dashed lines above 6 gpa show lattice parameters of optimal structure . the dotted lines present extrapolated values from the data below 6 gpa , namely those form monoclinic @xmath0 structure .
although optimal values ( dashed lines ) show the orthorhombic structure is stable above 6 gpa , the energy of the orthorhombic structure is very close to that of the monoclinic ( dotted lines ) structure .
since the @xmath2-o@xmath1 can be considered as a special case of @xmath0-o@xmath1 ( @xmath55=90@xmath56 ) , we illustrate the total energy as a function of @xmath55 and pressure .
figure 6 presents a pressure dependence of the cross section of the total energy from 4 gpa to 7 gpa . in the figure ,
lattice constants @xmath27 ,
@xmath6 and the volume of the unit cell are fixed to optimal values presented by the solid and dashed lines in figure 3(a ) . therefore the independent parameter is only @xmath55 at each pressure . at the low pressure range , the total energy takes minima at two values of @xmath55 , which give equivalent monoclinic @xmath0 structures , and the orthorhombic structure ( @xmath57 ) is unstable . with adopted constraints of the crystal structure and geometric configurations of molecules ,
it can be derived analytically that a first derivative of the total energy with respect to @xmath55 is zero at @xmath57 ( orthorhombic structure ) at each pressure , i.e. , the total energy takes a maximum or minimum values at the @xmath2 structure @xcite . with increasing pressure ,
the @xmath2 structure becomes stable at around 6 gpa in place of the @xmath0 structure , the energy difference between these structure is however less than 0.5 mev ( @xmath10 5 k ) even at 7 gpa .
unfortunately , in the precision of present calculations , the energy difference may be too small to decide which structure is stable .
we obtained however an important result that the monoclinic - angle - dependence of the total energy is very small in this pressure range .
et al . _ and gorelli _ et al . _
proposed different structures in this pressure range .
it is reasonable because the experiments were performed at temperatures higher than the energy difference we obtained .
the very small energy difference in this pressure range implies that several structures which have various monoclinic angles can be allowed at finite temperature .
et al . _ performed raman scattering experiments and observed peaks above 5 gpa @xcite . although the origin of the peaks is unresolved , they indicated the existence of the complicated mixed phase at this pressure range .
further investigations are desired to determine the origin of the unknown peaks . at high temperatures
, many experiments demonstrated the stability of orthorhombic @xmath2-o@xmath1 .
the stability of @xmath2-o@xmath1 at high temperature is understood from figure 6 . as shown in the figure , the energy surface forms a double - well or parabolic shape . in either case
, they are symmetric with respect to @xmath55 around @xmath57 and the energy barrier separating the wells are very low .
therefore the expectation value of @xmath55 becomes 90@xmath56 even at low pressure because of the effect of thermal excitations .
akahama _ et al .
_ reported abrupt changes of lattice parameters at the phase boundary of the @xmath0 and @xmath7 phases . in the present calculations , however , corresponding change of lattice parameters are not obtained up to 12 gpa .
this may be caused by constraints on the crystal and/or magnetic structure in our calculations .
recent neutron diffraction experiment shows a disappearance of the magnetic long - range ordering in the @xmath7 phase , but it does not exclude the short - range order @xcite .
intermolecular potentials adaptable to other magnetic configurations may be required to obtain the @xmath7 structure .
we investigated crystal structures of solid oxygen under pressure with _ ab - initio _ pair - potentials taking account of the difference of spin states . under 6 gpa ,
the obtained structure is monoclinic @xmath0 and calculated lattice parameters agree well with the results of the x - ray diffraction experiment .
the origin of the distortion of the basal plane and that of the anisotropic lattice compressibility of @xmath0-o@xmath1 are clearly demonstrated that they are caused by the existence of two kinds of intermolecular interaction , namely magnetic interactions .
the crystal structure under the pressure range from 6 to 7 gpa is discussed using the total energy surface projected on the @xmath58-@xmath55 space .
although orthorhombic @xmath2-o@xmath1 is obtained as the stable structure , the energy difference between @xmath0- and @xmath2-o@xmath1 is less than 5 k in this pressure range .
this implies a possibility of a mixed phase including some structures which have various monoclinic angles .
it seems to be consistent with recent x - ray diffraction and raman experiments .
we are grateful to y. akahama for stimulating discussions and for providing us with experimental data .
we would like to thank y. ishii , h. koizumi and y. mita for useful discussions and suggestions .
numerical calculations were partially carried out using the computer facilities at the research center for computational science , okazaki national research institutes . figure 1 : : ( a ) monoclinic unit cell of @xmath0-o@xmath1 .
the unit cell includes two oxygen molecules . at @xmath59 ,
the unit cell becomes an orthorhombic @xmath60 structure .
( b ) the basal plane of @xmath0-o@xmath1 .
oxygen molecules are denoted by circles and molecular axes are perpendicular to the @xmath13 plane .
arrows represent magnetic moment directions . figure 2 : : a schematic view of coordinates for pair - potentials .
the vector @xmath61 is the relative position of molecules .
figure 3 : : pressure dependences of ( a ) lattice constants and monoclinic angles ( b ) @xmath4 and ( c ) @xmath55 up to 7 gpa .
calculated values are denoted by solid lines and experimental values are represented by closed circles , triangles and diamonds .
see text regarding the dashed and dotted lines .
figure 4 : : a model of the @xmath13 plane .
circles denote oxygen molecules , and arrows represent magnetic moment directions .
the unit cell includes four af pairs and six f pairs .
two f pairs in which intermolecular distances are 2@xmath38 is not taken into account in the discussion .
figure 5 : : the optimized values of ( @xmath37 ) and pair - potentials .
these pair - potentials correspond to @xmath36 and @xmath35 , which are defined in previous section .
arrows denote the optimal values of ( @xmath37 ) .
note that the values of @xmath6 and @xmath38 are at almost minimum of pair - potentials .
figure 6 : : the contour plot of the total energy as a function of the pressure and @xmath55 in the pressure range 4 - 7 gpa . the volume of the unit cell and lattice constants @xmath27 and
@xmath6 are fixed at optimal values at each pressure , which are denoted by solid and dashed lines in figure 3 .
the difference in the energy between the monoclinic @xmath0 and orthorhombic @xmath2 structures is very small at around 6 gpa .
contours increase by 2 mev . | the crystal structure of solid oxygen at low temperatures and at pressures up to 7 gpa is studied by theoretical calculations . in the calculations ,
the adiabatic potential of the crystal is approximated by a superposition of pair - potentials between oxygen molecules calculated by an _ ab - initio _ method .
the monoclinic @xmath0 structure is stable up to 6 gpa and calculated lattice parameters agree well with experiments .
the origin of a distortion and that of an anisotropic lattice compressibility of the basal plane of @xmath0-o@xmath1 are clearly demonstrated . in the pressure range from 6 to 7 gpa , two kinds of structures are proposed by x - ray diffraction experiments : the @xmath0 and orthorhombic @xmath2 structures .
it is found that the energy difference between these structures becomes very small in this pressure range .
the relation between this trend and the incompatible results of x - ray diffraction experiments is discussed . |
this paper aims to generalize and extend some results in [ kw ] and [ w ] about deconstruction of cosmic microwave background radiation ( cmbr ) into multipole vectors and explain these results to a mathematically inclined audience .
our exposition , to a degree , is self - sufficient .
cosmologists who study the cmbr routinely decompose it into a sum of spherical harmonics
the eigenfunctions of the laplace operator on the @xmath15-sphere .
the summand that corresponds to the eigenvalue @xmath16 of the laplace operator is called a @xmath1-th _ spherical harmonic_. this decomposition helps to analyze the correlations between the magnitude and geometry of various harmonics .
the ultimate goal of these investigations is to understand the geometry of the visible universe and the physics of the big bang .
the data obtained by wilkinson microwave anisotropy probe ( wmap ) reveal puzzling correlations between the low-@xmath1 portion of the harmonic decomposition of the cosmic microwave background radiation [ b ] , [ chs ] , [ toh ] , [ ehgl ] . in the case of quadrupole - octopole ( @xmath17 and @xmath18 ) alignment , simulations based on our fast algorithm [ kw ] , [ w ]
show the alignment to be unusual at the 98.7 percent level .
no explanation is yet known for these strange results .
_ multipole vectors _ provide a convenient means to study the isotropic properties of cmbr .
in particular , they help to quantify the planarity of a given multipole , as well as to compare the alignment of two different multipoles [ chs ] .
jeff weeks and i , coming from a background in pure mathematics , were unable to decipher the formalism and terminology of [ chs ] and chose instead to re - create the multipole vector concept from scratch ( see [ kw ] ) .
later on , as we read the papers [ l ] and [ d ] , we realized that we managed to reinvent the wheel twice : our key algebraic lemma turns out to be a classical theorem of sylvester [ s ] , [ s1 ] , while the application of that lemma to harmonic polynomials on sphere implicitly was known to maxwell [ m ] .
the real - valued spherical harmonics of order @xmath1 are precisely the homogeneous harmonic polynomials of degree @xmath1 in the variables @xmath19 , @xmath20 and @xmath6 is the polynomial @xmath21 , up to normalization . ] .
thus , we sought to understand the multipole vectors of copi , huterer and starkman [ chs ] from a polynomial point of view .
as in [ ch ] , this calls for using some elementary tools from the field of algebraic geometry .
translated to the language of polynomials , [ chs ] s motivating goal was to factor every homogeneous harmonic polynomial @xmath8 of degree @xmath1 on the sphere into a product of linear factors @xmath22 such a factorization is impossible in general , as [ chs ] implicitly acknowledge by their introduction of suitable error terms .
the correct statement is given by the sylvester theorem [ s ] , [ s1 ] : by the authors . ]
every real homogeneous polynomial @xmath8 of degree @xmath1 in @xmath19 , @xmath20 and @xmath6 may be written as @xmath23 where the remainder term @xmath24 is a homogeneous polynomial of degree @xmath25 .
the decomposition is unique up to reordering and rescaling of the linear factors in the product . for a _ harmonic _ homogeneous @xmath8
, a somewhat similar representation was discovered by james clerk maxwell in his 1873 _ treatise on electricity and magnetism _ [ m ] .
its relation to algebraic geometry in general , and to the sylvester theorem , in particular , is well - explained by courant and hilbert in [ ch ] vii , sec .
we will describe the maxwell representation later in the paper .
sylvester s theorem 1 ( which is just an application of bzout s theorem ) has a pleasing corollary : @xmath26 $ ] when restricted to a unit sphere @xmath2 , every real polynomial @xmath8 in @xmath5 , and @xmath6 of degree @xmath1 can be written in the form @xmath27 the decomposition is unique up to reordering and rescaling of the linear factors in each of the products .
in particular , one can pick each @xmath28 to be a unit vector and , for @xmath29 , each @xmath30 .
it is well - known that spherical harmonics form a basis in the space @xmath31 of @xmath32-integrable functions on a unit sphere ( see [ ch ] ) . as a result
, an infinite maxwell - type decomposition is available for any function @xmath33 ( cf .
[ l ] , formula @xmath34 ) .
any real function @xmath35 has a representation in the form @xmath36 where the series converges in the @xmath32-norm , and the mutually orthogonal polynomials in the figure brackets are among the @xmath1-th spherical harmonics .
this representation is unique , up to reordering and rescaling of the linear factors in each of the products . in the space of real continuous functions @xmath37 on @xmath2 there is a dense ( in the @xmath38-norm ) and @xmath39-invariant subset @xmath40 comprised of functions that admit a representation in the form of a series @xmath41 that uniformly converges on the sphere .
each functions from @xmath40 admits a canonical extension into the interior of the unit ball @xmath42 where it produces a real analytic function .
this theorem , claiming a form of stability of decomposition @xmath43 under the polynomial approximations , is a useful tool for analyzing patterns in the spherical sky [ w ] . at the first glance
, the sphere seems to occupy a special place in these results .
however , what is really important , is the _
quadratic nature _ of the surface .
this paper is concerned with similar decompositions of functions on other quadratic algebraic skies ( real and complex ) .
for instance , it is natural to wonder : can one deconstruct in a similar fashion polynomial or real analytic functions on a hyperboloid ?
the short answer is yes , one can " , but the uniqueness of the representation is lost .
the focal point of this paper is to describe rich and interesting algebro - geometrical structures and topology of modular spaces of multipoles . in the process , we bring under the same roof a variety of mathematical techniques and constructions that belong to the fields of algebraic geometry , singularity theory , algebraic topology , harmonic analysis , and the polynomial approximation theory .
none of our techniques is very advanced , but their natural appearance within the context of studying quadratic surfaces is pleasing and somewhat surprising ...
the reader can follow two distinct treads in the core of the paper .
the first unifies results that are linear in nature and are based on the classical theory of harmonic analysis on quadratic surfaces .
these results are valid for quadratic hypersurfaces _ in any dimension_. the second tread winds through the results that require methods of algebraic geometry .
these results are uniquely _ three - dimensional _ , that is , applicable only to quadratic _
we denote by @xmath44 $ ] the ring of complex polynomials in the variables @xmath45 and @xmath6 .
let @xmath12 be an irreducible quadratic form over the complex numbers @xmath46 . from a strictly algebraic perspective , one can interpret this section as describing the ways in which the quotient ring @xmath44 / \langle q - \lambda \rangle$ ] , @xmath47 , fails to be a unique factorization domain .
however , the flavor of our approach to this problem is more geometrical and combinatorial
. the reader will be well - adviced to ignore for a while the asterisks in our notations : they are there to distinguish between vector spaces and their duals .
let @xmath48 be a complex algebraic surface in @xmath49 .
at the same time , @xmath50 gives rise to a complex projective curve @xmath51 , where @xmath52 stands for the projectivization of the @xmath18-space with coordinates @xmath5 , and @xmath6 . as in theorem 2 , we aim to decompose any polynomial @xmath0 restricted to @xmath53 as an economic " sum of products of homogeneous linear polynomials .
first , consider the case of an homogeneous @xmath8 and the corresponding complex projective quadratic curve @xmath54 in @xmath55 it generates .
denote by @xmath56 the intersection @xmath57 .
we assume that @xmath54 and @xmath58 do not share a common component , so that @xmath56 is a finite set . when @xmath56 is a complete intersection , it consists of exactly @xmath59 points , where @xmath60 .
let @xmath12 be an homogeneous irreducible quadratic polynomial , and @xmath0 an homogeneous polynomial of degree @xmath1 .
assume that @xmath56 is a complete intersection .
a _ parcelling _ of the set @xmath56 is comprized of a number of subsets @xmath61 so that : * @xmath62 * @xmath63 , provided @xmath64 * the cardinality of each subset @xmath65 is equal to 2 . when @xmath66 is irreducible and @xmath8 is not divisible by @xmath66 , the intersection set @xmath56 is equipped with a function @xmath67 which assigns to each point @xmath68 its multiplicity @xmath69 .
we denote by @xmath70 the set of non - negative integers and by @xmath71 the set of positive integers .
let @xmath72 be a finite set equipped with a multiplicity function @xmath73 whose @xmath74-norm @xmath75 is @xmath59 .
a _ generalized _ _ parcelling _ of @xmath76 is a collection of functions @xmath77 , such that * @xmath78 * @xmath79 of course , each parcelling of @xmath56 is also a generalized parcelling , where the roles of the functions @xmath80 are played by the characteristic functions of the parcels @xmath65 . in the case of @xmath81 ,
the following lemma is due to sylvester [ s ] , [ s1 ] .
let @xmath12 be an irreducible homogeneous quadratic polynomial and let @xmath0 be a homogeneous polynomial of degree @xmath1 which is not divisible by @xmath66 .
consider a generalized parcelling @xmath82 of the multiplicity function @xmath83 .
then the homogeneous polynomial @xmath0 admits a representation of the form @xmath84 where @xmath85 denotes a linear homogeneous polynomial that vanishes at each point @xmath86 with the multiplicity @xmath87 . * proof * there exist a linear polynomial @xmath88 and a corresponding line @xmath89 such that the multiplicity of @xmath90 at each point @xmath91 equals @xmath87 .
when @xmath92 , the line must be tangent to @xmath93 at @xmath94 .
when the support of @xmath80the parcel @xmath95)is comprised of two points , the line @xmath96 is chosen to contain @xmath65 .
let us compare the restrictions of @xmath8 and @xmath97 to the curve @xmath93 .
both polynomials are of the same degree @xmath1 .
moreover , the curve @xmath98 intersects with @xmath93 at the points of @xmath56 , where it realizes the multiplicities prescribed by the function @xmath67 .
it follows that the restrictions of @xmath8 and @xmath97 to @xmath93 are proportional , that is , for an appropriate choice of scalar @xmath99 , @xmath100 .
just take @xmath101 where @xmath102 . with this choice , the curves @xmath103 and @xmath93 intersect so that the total multiplicity of the intersection is at least @xmath104 . by the bzout theorem ,
this is possible only when @xmath105 . employing the irreducibility of @xmath66
, we get that @xmath106 must be divisible by @xmath66 .
this completes the proof .
any effective divisor @xmath108 on the complex quadratic surface @xmath109 that is the zero set of a homogeneous polynomial @xmath0 can be represented as a sum of lines @xmath110 .
@xmath107 in general , the representation ( 6 ) of @xmath8 depends on a generalized parcelling , and thus , is not unique .
we denote by @xmath111 the complex vector space of homogeneous polynomials in @xmath5 , and @xmath6 of degree @xmath1 .
its dimension is @xmath112 .
consider a vector subspace @xmath113 comprised of polynomials divisible by @xmath66 .
there is a canonical imbedding @xmath114 whose image is @xmath115 .
it is produced by multiplying polynomials of degree @xmath25 by @xmath66 .
thus , any space @xmath111 is equipped with a natural filtration @xmath116 by subspaces of polynomials divisible by various powers @xmath117 of @xmath66 .
we equip each @xmath111 with an hermitian inner product . for various @xmath1 s , we insist that these inner products are synchronized by the requirement : all the imbeddings @xmath118 must be isometries .
denote by @xmath119 the subspace of @xmath111 orthogonal to @xmath115 .
it can be identified with the quotient space @xmath120the @xmath1-graded part of the algebra of regular functions on the complex cone @xmath121 .
the dimension of @xmath119 is equal to @xmath122 . for a real form @xmath66 , similar spaces based on real homogeneous polynomials of a degree @xmath1 make sense .
we denote them @xmath123 , and @xmath124 . when @xmath81 , there is a classical interpretation for the quotient @xmath125 .
it can be identified with the set of _ harmonic _ homogeneous polynomials of degree @xmath1 ( cf .
[ ch ] and [ m ] ) . in other words ,
the kernel @xmath126 of the laplace operator @xmath127 is complementary to @xmath128 in @xmath129 .
this leads to a decomposition similar to the one in @xmath130 ( cf . [
sh , theorem 22.2 ] , [ l ] , and [ w ] ) : @xmath131 the direct summands in @xmath132 are orthogonal with respect to the inner product in @xmath129 defined by the formula @xmath133 .
the measure @xmath134 on the sphere @xmath2 is the standard one . moreover , according to maxwell , any polynomial @xmath135 admits a beautiful representation of the form @xmath136 where @xmath137 , @xmath138 are some vectors , . ] and @xmath139 stands for the directional derivative operator .
physics behind maxwell s representation is quite transparent : @xmath140 is a potential of a single electrical charge , @xmath141 is a potential of a virtual " ( that is , very small ) dipole formed by two opposite charges , @xmath142 is a potential of a virtual " quadropole a close pair of @xmath143-oriented dipoles merging along the direction of @xmath144 , @xmath145 is a potential of a virtual " octopole , and so on ... note that , the laplace operator @xmath146 can act on complex polynomials as well . moreover ,
if @xmath147 for some analytic function @xmath37 , then its real and imaginary parts are also harmonic : @xmath148 and @xmath149 . for any invertible complex @xmath150-matrix @xmath151 , consider a symmetric matrix @xmath152 and the corresponding quadratic form @xmath153 , where @xmath154 , @xmath155 stands for the inner product @xmath156 in @xmath157 , and the upper script @xmath158 denotes the matrix transposition . employing @xmath66 , we can form a second order differential operator @xmath159 acting on holomorphic functions @xmath37 in the complex variables @xmath160 . here
@xmath161 denote elements of the inverse matrix @xmath162 .
formally , @xmath163\cdot b^{-1 } \cdot [ \vec{\partial}]^t$ ] where @xmath164 : = ( \partial_1 , \partial_2 , \partial_3 ) : = ( \partial_x , \partial_y , \partial_z)$ ] .
given a non - degenerated quadratic form @xmath66 , there is a change of complex coordinates @xmath165 that reduces it to the canonical form @xmath166 .
consider the complex laplace operator @xmath167 in the new coordinates @xmath168 .
then , for any holomorphic function @xmath169 , we have @xmath170 ( ( x , y , z)\cdot a ) = \delta_q [ f ( ( x , y , z)\cdot a)]\ ] ] thus , there is a 1-to-1 correspondence between harmonic homogeneous polynomials @xmath171 of degree @xmath1 and degree @xmath1 homogeneous polynomial solutions @xmath172 of the equation @xmath173 let us examine the intersection @xmath174 .
for any homogeneous polynomial @xmath158 of degree @xmath25 , we get @xmath175 . therefore , if @xmath176 , then @xmath158 must be divisible by @xmath177 ( note that @xmath178 implies @xmath179 ) . put @xmath180 and @xmath181 .
the equation @xmath182 is equivalent to the equation @xmath183\cdot t_1 $ ] .
again , it follows that @xmath184 must be divisible by @xmath177 .
continuing inductively this kind of reasoning , we see that @xmath185 . on the other hand
, one can verify that @xmath186 + dim [ v_{q'}(d ) ] = dim [ v(d)]$ ] .
thus , @xmath187 .
a polynomial @xmath188 is divisible by @xmath189 , if and only if , @xmath190 is divisible by @xmath12 .
therefore , @xmath191 as well .
let @xmath192 denote a subgroup of the general linear group @xmath193 that preserves the quadratic form @xmath66 .
a matrix @xmath194 if and only if @xmath195 . the natural @xmath192-action on the space of @xmath5 , and @xmath6-variables induces an action on the polynomial space @xmath111 .
evidently , @xmath115 is invariant under this action . on the other hand , for any polynomial @xmath0 , @xmath196 = [ \delta_{\tilde q } p]((x , y , z)\cdot u)),\ ] ] where the operator @xmath197\cdot u^t b^{-1 } u \cdot [ \vec{\partial}]^t$ ]
. since @xmath195 , we get @xmath198 . by a simple algebraic trick
, it follows that @xmath199 . as a result ,
both the quadratic form @xmath66 and the kernel @xmath200 are invariant under the @xmath192-action .
consider maxwell s representation ( 8) of a real homogeneous harmonic polynomial @xmath188 of degree @xmath1 .
it gives rise to a map @xmath201 that takes the sets of vectors @xmath202 to elements of @xmath203 .
the map @xmath201 is evidently linear in each of the @xmath204 s , and thus it is a real polynomial map with a vector space of real dimension @xmath104 for its target . by [ ch ] , [ l ] and [ w ] , @xmath201 is onto the vector space @xmath203 .
therefore , the complexification @xmath205 of @xmath201 must be also onto the vector space @xmath206 ( indeed , the image of @xmath205 in @xmath207 is a complex algebraic set containing a totally real vector subspace @xmath203 of a maximal dimension ) .
in other words , formula ( 8) must be valid for any complex homogeneous harmonic polynomial @xmath188 of degree @xmath1 and appropriate complex vectors @xmath208 .
in fact , each vector @xmath209 .
formula ( 8) describes any homogeneous complex polynomial solution of the equation @xmath210 in the @xmath168 coordinates . translating them back to the @xmath211-coordinates with the help of the identity @xmath212(\mathbf x\cdot a ) = \nabla_{\mathbf v ' \cdot a^{-1 } } [ f'(\mathbf x \cdot a)]$ ] leads to a formula @xmath213 below .
thus , we have established the following proposition : the decomposition @xmath214 holds for any non - degenerated complex quadratic form @xmath66 .
it is invariant under the natural @xmath192-action on @xmath111 .
moreover , for any @xmath215 of degree @xmath1 , there exist vectors @xmath216 so that the generalized maxwell formula @xmath217 is valid .
the real part of the lhs of @xmath213 generates all real homogeneous polynomial solutions @xmath8 of the equation @xmath218 . in particular , with @xmath219 ,
any degree @xmath1 homogeneous complex polynomial solution @xmath8 of the wave equation @xmath220 is given by the formula @xmath213 .
@xmath107 formula @xmath213 might pose a slight challenge : after all , square roots are multivalued analytic functions . however , the @xmath221-ambiguities in picking a single - valued branch cancel each other .
we shall see that , even up to reordering and rescaling of the multipole vectors @xmath222 , the complex maxwell representation @xmath213 of a given @xmath66-harmonic " @xmath8 is not unique .
* in fact , by a similar argument , the decomposition ( 9 ) is available for homogeneous polynomials in _ any number of variables _ and for any non - degenerated quadratic form @xmath66 ( see [ sh ] , theorem 22.2 , for the proof ) . at the same time
, the representation ( 10 ) seems to be a 3-dimensional phenomenon .
it looks like that not any @xmath66-harmonic polynomial in @xmath223 variables can be expressed in terms of @xmath1 directional derivatives sequentially applied to the @xmath66-harmonic potential @xmath224 .
this seems to be related to the failure of the sylvester - type formula ( 6 ) for @xmath223 .
when i raised this issue with michael shubin , he proposed a nice conjecture in the flavor of maxwell s representation ( although , not a direct generalization of it ) . in the maxwell representation ,
one employs a differential operator which is a _ monomial _ in @xmath1 directional derivatives , while in the conjecture below one invokes a differential operator which is a _ polynomial _ in @xmath1 directional derivatives .
* shubin s conjecture * ( _ an @xmath225-dimensional variation on the theme of maxwell s representation _ ) .
let @xmath226 .
for any real homogeneous and harmonic polynomial @xmath8 of degree @xmath1 in @xmath225 variables @xmath227 , there exists a unique real homogeneous and harmonic polynomial @xmath228 also in @xmath225 variables such that @xmath229,\ ] ] where @xmath230 , and the differential operator @xmath231 being applied to the potential function @xmath232 .
moreover , the polynomials @xmath8 and @xmath228 are proportional with the coefficient of proportionality depending only on @xmath1 and @xmath225 .
@xmath107 given a vector space @xmath233 , we denote by @xmath234 the space @xmath235 .
we consider the @xmath18-space @xmath236 of linear forms @xmath237 ( which we identify with the space of their coefficients @xmath238 ) and the @xmath18-space @xmath239 with the coordinates @xmath240 .
this calls for a distinction in notations for their projectivizations : @xmath241 and @xmath52 . hence
, each point @xmath242 determines a point @xmath243 $ ] in @xmath241 and a complex line @xmath244 in @xmath52 the zero set of the form @xmath245 .
now , let us return to the decomposition @xmath130 .
each polynomial @xmath88 from @xmath130 can be viewed as a vector in @xmath246 .
therefore , a given homogeneous polynomial @xmath8 of degree @xmath1 which is not divisible by @xmath66 , together with the appropriate generalized parcelling , produces a collection of non - zero vectors @xmath247 .
we shall call this unordered set of vectors @xmath248 _ the leading multipole vectors of _ @xmath8 with respect to the corresponding generalized parcelling of @xmath249 . for a given generalized
parcelling @xmath82 of @xmath249 , the subordinate leading multipole vectors @xmath250 are unique up to reordering and rescaling of the @xmath88 s ( equivalently , the polynomial @xmath24 in @xmath130 is unique ) .
* proof*we use the same notations as in the proof of lemma 1 . by an argument as in that proof , any linear polynomial @xmath251 which defines a line @xmath252 with the property @xmath253 ( the points of intersection have multiplicities prescribed by @xmath80 ) is of the form @xmath254 . here
@xmath255 and @xmath88 is a preferred polynomial corresponding to a vector @xmath256 . indeed , as in the proof of lemma @xmath257 , for an appropriate @xmath99 , by the bzout theorem , @xmath258 must be divisible by @xmath66 .
@xmath107 as a result , the multipole is well defined by the parcelling of @xmath67 modulo some rescaling .
clearly , one can always replace each @xmath88 in @xmath130 with @xmath259 as long as @xmath260 .
next , we analyze the exact meaning of the reordering and rescaling " ambiguity .
let @xmath261 be an abelian subgroup of @xmath262 formed by vectors @xmath263 subject to the restriction @xmath260 .
its rank is @xmath264 .
let @xmath265 stand for the symmetric group in @xmath266 symbols @xmath267 .
we denote by @xmath268 an extension @xmath269 of @xmath265 by @xmath261 .
this group is generated by the obvious actions of @xmath265 and @xmath261 on @xmath270 .
we introduce an orbit - space @xmath271^k/ \sigma_k\end{aligned}\ ] ] whose points encode the products of linear forms as in the decomposition @xmath130 . here
the group @xmath272 acts on the products of spaces by permuting them and by rescaling their vectors .
its subgroup @xmath273 acts freely . because of the uniqueness of the prime factorization in the polynomial ring @xmath44 $ ] , the space @xmath274 provides us with a @xmath257-to-@xmath257 parametrization of the space of homogeneous degree @xmath275 polynomials that are products of linear forms . as an abstract space , ( 11 ) can be expressed as @xmath276^k \big\}\big/\sigma_k \end{aligned}\ ] ] the elements of orbit - space @xmath274 will be called @xmath275-_poles _ , or simply , _
multipoles_. we introduce groups @xmath277 in a manner similar to the introduction of the groups @xmath272 .
the group @xmath277 is an extension of the permutation group @xmath278 by the group @xmath279 .
thus , @xmath280 .
therefore , the space @xmath274 in @xmath281 fibers over the projective variety @xmath282 \approx sym^k ( \mathbb cp^2)\end{aligned}\ ] ] with the fiber @xmath283 . here
@xmath284 denotes the @xmath285-th symmetric power of a space @xmath286 .
the natural map @xmath287 is a principle @xmath283-fibration which gives rise to a line bundle @xmath288 .
by shrinking its zero section to a point , we form a quotient space @xmath289 \ , / \mathcal b(k ) .
\end{aligned}\ ] ] it differs from @xmath274 by a single new point @xmath290a point which will represent the zero multipole .
evidently , @xmath291 is a contractible space . given a collection of vector spaces @xmath292 ,
their _ wedge _ product @xmath293 ( not to be mixed with the exterior product ! ) is the quotient of the cartesian product @xmath294 by the subspace comprised of sequences @xmath295 such that at least one vector from the sequence is zero .
thus , topologically , @xmath296/ \sigma_k.\end{aligned}\ ] ] lemma 3 leads to the following proposition . consider an homogeneous polynomial @xmath0 of degree @xmath1 which is not divisible by an irreducible homogeneous quadratic polynomial @xmath12 .
then any generalized parcelling @xmath82 of the multiplicity function @xmath249 uniquely determines a multipole in the space @xmath297 introduced in @xmath298 or @xmath299 . @xmath107
the space of multipoles has singularities .
its singular set @xmath300 arises from the sets points in @xmath301^k$ ] fixed by various non - trivial subgroups of @xmath272 .
these subgroups all are the conjugates ( in @xmath272 ) of certain non - trivial subgroups of @xmath278 ( recall that @xmath302 acts freely ) .
the partially ordered set of the orbit - types give rise to a natural stratification of the multipole space @xmath274 .
the space @xmath300 is of complex codimension two in @xmath274 .
its top strata corresponds to transpositions from @xmath278 .
thus , a generic point from @xmath300 has a normal slice in @xmath274 which is diffeomorphic to a cone over the real projective space @xmath303 .
the larger stabilizers of vectors from the space @xmath304 of the obvious @xmath278-representation correspond to smaller strata of more complex geometry .
evidently , @xmath300 is invariant under the diagonal action of @xmath305 .
these observations are summarized in the lemma below .
the singular set @xmath306 is of codimension two .
it is invariant under the @xmath283-action .
therefore , it is a principle @xmath283-fibration over the singular set @xmath307 .
a generic point of @xmath300 has @xmath308 for its normal link .
points of @xmath300 correspond to completely factorable polynomials @xmath309 with at least two proportional linear factors ( in other words , to weighted collections of lines in @xmath52 that contain at least one line of multiplicity greater than one ) .
@xmath107 let @xmath310 denote the variety of homogeneous polynomials of degree @xmath1 in @xmath5 , and @xmath6 that are products of linear forms
. given a multipole @xmath311 one can construct the corresponding completely factorable polynomial @xmath312 .
note that , due to the uniqueness of the prime factorization in the polynomial ring and in view of our definition of multipoles , the correspondence @xmath313 gives rise to a @xmath257-to-@xmath257 map @xmath314 .
consider a
vite - type " algebraic map @xmath315 where @xmath316 is induced by restricting polynomials in @xmath5 , and @xmath6 to the surface @xmath121 .
the symbol @xmath317 denotes the variety of homogeneous polynomial functions on the surface @xmath318 that also decompose into products of linear forms . due to formula @xmath130 in lemma 1 , any non - zero homogeneous polynomial on the surface
@xmath318 admits a linear factorization .
hence , @xmath319 is _ onto _ and @xmath320 can be identified with the space @xmath321^\circ \approx v_q^{\perp}(d)^\circ$ ] .
the map @xmath319 , extends to an algebraic map @xmath322 defined on the space @xmath323 of the line bundle @xmath324 .
it sends the zero section @xmath325 of @xmath324 to the zero vector @xmath326 and each fiber of @xmath324 isomorphically to a line in @xmath327 passing through the origin .
in fact , @xmath328 .
evidently , @xmath329 gives rise to a continuous map @xmath330 it turns out that @xmath331 has _
finite _ fibers .
we need some combinatorial constructions which will help us to prove this claim and to describe the cardinality of the @xmath331-fibers . with every natural @xmath1 we associate an integer @xmath332 that counts the number of distinct parcellings in a finite set of cardinality @xmath59 .
any parcelling is obtained by breaking a set @xmath72 of cardinality @xmath59 into disjoint subsets of cardinality 2 .
thus , @xmath333 is the number of possible handshakes among a company of @xmath59 friends .
there is an alternative way to compute this number : consider the standard action of the permutation group @xmath334 on the set of @xmath59 elements .
the action induces a transitive action on the set of all parcellings . under this action , the subgroup @xmath335 that preserves
the parcelling @xmath336 is an extension @xmath337 of the permutation group @xmath338 that acts naturally on the pairs by the group @xmath339 that exchanges the elements in each pair . as a result
, we get an identity @xmath340 as we deform a polynomial @xmath8 into a polynomial @xmath341 , two or more points in @xmath56 can merge into a single point of @xmath342 .
its multiplicity is equal to the sum of multiplicities of the points forming the merging group . through this process
, any generalized parcelling @xmath94 of the original @xmath343 gives rise to a new and unique generalized parcelling @xmath344 of @xmath345 .
thus , we can define a partial order in the set of all generalized parcellings of effective divisors on @xmath93 of degree @xmath59 by setting @xmath346 and @xmath347 ( see figures 1 and 3 ) . in the same spirit ,
let @xmath348 stand for the number of distinct generalized parcellings of a function @xmath73 ( recall that @xmath349 ) on a finite set @xmath72 . unless @xmath67 is identically @xmath257 and @xmath350 , @xmath351 .
when two intersection points merge , the number of generalized parcellings drops : there are distinct original parcellings that become indistinguishable after the merge ( see the left diagram in figure 1 ) . for example , when two simple points in a complete intersection merge , the number of generalized parcellings changes from @xmath332 to @xmath352/2 = [ \kappa(d ) + \kappa(d - 2)]/2 $ ] , that is , it drops by @xmath353/2 $ ] . in general , @xmath346 implies @xmath354 . for a generic @xmath355 the set @xmath356 , as well as all the parcels @xmath357 defined by the linear factors @xmath88 ,
are complete intersections .
for such an @xmath245 , the number of multipoles in @xmath358 is the number @xmath332 of distinct parcellings in the set @xmath356 . for any @xmath355
, the cardinality of @xmath358 ( equivalently , of @xmath359 ) equals to the the number of generalized parcellings of the multiplicity function @xmath360 .
indeed , assume that two completely factorable polynomials @xmath361 coincide when restricted to the surface @xmath318 .
then @xmath362 must be divisible by @xmath66 .
therefore , @xmath363 , moreover , the two multiplicities of each point in the intersection ( defined by the curves @xmath244 and @xmath364 ) must be equal as well .
hence , @xmath244 and @xmath364 define two parcellings of the same multiplicity function @xmath67 on the intersection set .
let @xmath365 be topological spaces and @xmath366 a continuous map with finite fibers .
for a while , the _ ramification set _
@xmath367 of @xmath37 is understood as the set @xmath368 such that , for any open neighborhood @xmath369 of @xmath370 , the cardinality of the fibers @xmath371 is not constant .
each time @xmath372 produces in @xmath52 a union of lines with at least one pair of lines sharing its intersection point with the curve @xmath93 , the point @xmath373 . moreover , as figure 1 demonstrates , this change in cardinality of fibers occurs due to their bifurcations in the vicinity of @xmath245 . as a result , points in @xmath374 give rise to effective divisors of degree @xmath59 on the curve @xmath93 with at least one point in their support being of multiplicity greater than one . on the other hand ,
any such divisor can be generated by intersecting a weighted set of lines with @xmath93 ( just use any generalized parcelling ) .
note that any line tangent to @xmath93 also contributes a point of multiplicity two .
however , the lines tangent to @xmath93 are not contributing to the bifurcation of the @xmath316-fibers ( see the right diagram in figure 1 ) .
therefore , in order to conclude that any divisor on @xmath93 with multiple points corresponds to a point of @xmath375 , we need to use generalized parcellings which favor pairs of lines that share their intersection with @xmath93 to a single tangent line ( both patterns produce an intersection point of multiplicity two as shown in figure 1 ) .
evidently , this can be done , provided @xmath376 .
the requirement that @xmath372 has a pair of linear forms vanishing at a point of @xmath318 , locally , is a single algebraic condition imposed on the coefficients of of the two forms .
therefore , taking closures , it picks a codimension one subvariety @xmath377 .
thus , @xmath378 is a subvariety of codimension one as well .
fortunately , since @xmath93 admits a rational parameterization by a map @xmath379 , the set @xmath380 can be described in terms of solvability of a system of two rather simple equations . in homogeneous coordinates
@xmath381 $ ] such a parameterization @xmath382 can be given by the formula @xmath383 ) = [ \alpha_0([u_0 : u_1 ] ) : \alpha_1([u_0 : u_1 ] ) : \alpha_2([u_0 : u_1])],\ ] ] where @xmath384 are some quadratic forms .
for example , for @xmath81 , @xmath385 the inverse of @xmath382 is produced by the central projection of @xmath93 onto a line in @xmath52 from a center located at @xmath93 . therefore , as _ abstract _ algebraic curves , @xmath93 and @xmath386 are isomorphic .
in fact , @xmath387 belongs to @xmath380 , if and only if , the system @xmath388 that guarantees an existence of a multiple zero for the polynomial @xmath389 in @xmath386 , has a non - trivial solution @xmath390 .
writing down explicitly the resultant of the two polynomials in the lhs of @xmath391 , and thus the equation of @xmath380 , seems to be cumbersome .
note that the euler identity @xmath392 explains the asymmetry of @xmath391 with respect to the variable @xmath393 : exchanging the roles of @xmath393 and @xmath394 leads to an equivalent system of equations .
recall that for a projective variety @xmath286 , the set of zero - dimensional , degree @xmath1 effective divisors is the projective variety @xmath395 ( see [ ch ] ) .
the map @xmath319 in formula @xmath396 induces a well - defined regular map of the varieties : @xmath397 this map is produced by realizing a given multipole @xmath398 , or rather its @xmath283-orbit @xmath399 , by a completely factorable polynomial @xmath400 and then forming the intersection set @xmath401 equipped with the appropriate multiplicities the divisor @xmath402 .
the map @xmath403 is onto : any effective divisor of degree @xmath59 on @xmath93 admits a generalized parcelling , and thus is generated by intersecting @xmath93 with a weighted collection of lines . since the number of generalized parcellings of @xmath401 is finite , the map @xmath403 has finite fibers .
all this can be seen from a different angle .
the divisor @xmath402 uniquely determines the proportionality class of the function @xmath404a point in @xmath405 .
we have seen that the @xmath283-equivariant map @xmath319 in @xmath396 is onto .
thus , @xmath406 is onto as well .
since @xmath93 is a rational curve ( topologically , a 2-sphere ) , there is a 1-to-1 correspondence between points of @xmath407 and of @xmath408 .
moreover , as abstract varieties , @xmath405 and of @xmath408 are isomorphic .
this isomorphism can be used to identify the maps @xmath403 and @xmath409 . to simplify our notations
, we will use the same symbol @xmath403 for both maps ; when there is a need to distinguish them , we will just indicate the relevant target space .
our combinatorial considerations imply that the locus @xmath410 consists of the effective divisors of degree @xmath59 on @xmath93 with at least one point in their support being of multiplicity at least two . while @xmath411 is also described by @xmath391 , the locus @xmath412 is the preimage of effective divisors of degree @xmath59 on @xmath93 with at least a pair of points in their support being of multiplicity at least two ( they can generate a double line ) , or at least one point being of multiplicity at least four ( it corresponds to a double line tangent to @xmath93 ) . employing lemma 4 ,
we have established the ramification set @xmath374 for the map @xmath413 is the set @xmath414 of complex codimension one , where @xmath372 defines on @xmath93 an effective divisor with at least one point in its support being of multiplicity at least two . in other words , @xmath415 is defined by the resultant of the two polynomials in the lhs of @xmath391 . the ramification set @xmath374 contains the @xmath319-image of the singular set @xmath416 .
this image is of codimension two in @xmath119 .
it can be identified with completely factorable homogeneous polynomials of degree @xmath1 on the the surface @xmath318 that have at least one pair of proportional linear factors .
@xmath107 note that the space @xmath72 of _ simple _ effective degree @xmath275 divisors on the curve @xmath93 is homeomorphic to the space of simple effective degree @xmath275 divisors on the sphere @xmath417 .
therefore , @xmath418 , where @xmath419 stands for the braid group in @xmath275 strings residing in the spherical shell @xmath420 $ ] . in particular , @xmath421 is a @xmath422-space .
let @xmath423 be the preimage of the subgroup @xmath424 in @xmath425 ( of order @xmath426 ) under the canonical epimorphism @xmath427 .
we call it _ the braid group in _ @xmath59 _ _ strings with coupling _ _ is not the braid group in @xmath59 strings colored with @xmath1 colors , each color marking a pair of strings ! ] .
it is a subgroup of index @xmath428 in the braid group @xmath429 .
recall that a covering of a @xmath13-space is again a @xmath430-space , where @xmath431 is an appropriate subgroup of @xmath14 .
since all our constructions are @xmath283-equivariant , lemmas 4 and 5 together with the arguments above lead to * the map @xmath432 is a finite ramified covering with a generic fiber of cardinality @xmath433 .
it is ramified over the subvariety @xmath410 whose points are the proportionality classes of homogeneous polynomials of degree @xmath1 on the quadratic surface @xmath121 that define there effective divisors of degree @xmath1 with at least one multiple line effective divisors with a point of multiplicity at least two . ] . *
the space @xmath434 is a @xmath13-space with the group @xmath14 being isomorphic to the braid group @xmath429 in @xmath59 strings in the spherical shell @xmath420 $ ] .
* as a result , @xmath435 is a @xmath436-space , where @xmath423 is the braid group in @xmath59 strings with coupling . *
the space @xmath437 also is @xmath13-space with the group @xmath14 being isomorphic to an extension of the infinite cyclic group @xmath438 by the group @xmath423 .
@xmath107 now , we will investigate the ramification locus of @xmath319 from a more refined point of view characteristic to the singularity theory .
first , we would like to understand when a non - zero vector @xmath398 from the tangent cone @xmath439 of @xmath440 at a point @xmath441 is parallel to the subspace @xmath115 , in other words , when @xmath439 contains a vector @xmath398 that is mapped to zero under the projection @xmath442 .
away from the singularity set @xmath443 , the @xmath316-image of such an @xmath245 belongs to a locus @xmath444 over which @xmath445 at some point in @xmath446 . by the implicit function theorem , the ramification locus @xmath447 for the map @xmath448 ( equivalently , for the map @xmath319 ) is contained in the union @xmath449 .
it can happen that at a singularity @xmath450 the tangent cone does not have vectors @xmath451 with the property @xmath452 , and still @xmath14 is ramified in the vicinity of @xmath453 .
for instance , consider the obvious projection @xmath14 of the real cone @xmath454 onto the @xmath455-plane : @xmath14 is ramified at the origin @xmath456 , but for any @xmath451 from the tangent cone at @xmath457 , @xmath458 .
any @xmath459 is of the form @xmath460 , where each @xmath461 is an appropriate linear form in @xmath462 , and @xmath463 or @xmath464 identically .
this limit is equal to the polynomial @xmath465 .
the vector @xmath8 at @xmath245 is parallel to the subspace @xmath466 if and only if the polynomial @xmath8 is divisible by @xmath66 . in other words ,
the vector @xmath8 at @xmath245 is parallel to @xmath466 if and only if the polynomial @xmath8 , being restricted to the curve @xmath93 , is identically zero .
thus , we are looking for the @xmath461 s subject to the constraint : the polynomial @xmath467 , being restricted to @xmath93 , is identically zero . for @xmath468 ,
this is equivalent to the constraint @xmath469 imposed on the rational functions @xmath470 : @xmath471
equation @xmath469 always have _ obvious _ solutions : @xmath472 , where @xmath473 and @xmath474 . these are exactly solutions that represent the zero tangent vector ( the tip @xmath245 of the tangent cone ) . indeed , put @xmath475 to conclude that @xmath476 if and only if @xmath474 .
so , the proper question is how to describe all the @xmath372 for which @xmath469 has a solution _ distinct _ from the set of obvious solutions @xmath472 with @xmath474 .
the images of such @xmath245 s under the projection @xmath477 will generate the locus @xmath444 over which the differential @xmath478 is not of the maximal rank @xmath479 . evaluating the lhs of equation @xmath469 at @xmath104 generic points residing in @xmath93 imposes linear constraints on @xmath480 variables the coefficients of the @xmath461 s .
if these constraints are independent , the solution space of the linear system must be of dimension @xmath481 , which is exactly the dimension of the space formed by the obvious solutions .
this indicates that , for a generic @xmath245 , we should not expect any non - obvious solutions .
lemma 6 below validates this guess .
let us denote by @xmath482 the quotient of the vector space of all solutions @xmath483 of @xmath469 by the subspace of obvious solutions , as defined above .
the correspondence @xmath484 gives rise to a new natural stratification of the space @xmath440 , and thus , of the space @xmath485 . in the new notations ,
@xmath486 if and only if @xmath487 .
we suspect that this stratification is consistent with , but cruder than the stratification induced by the orbit - types of the @xmath488-action .
the loci @xmath489 and @xmath490 coincide . as a result ,
the existence of a non - trivial solution for the system @xmath391 is equivalent to the existence of a non - obvious solution for the equation @xmath469 .
also , the locus @xmath491 . * proof .
* we notice that if two distinct lines , say @xmath492 and @xmath493 , share a point @xmath494 , then @xmath469 has a non - obvious solution @xmath495 .
indeed , inscribe in the quadratic curve @xmath93 any `` quadrilateral '' formed by the pair of lines @xmath496 together with a new pair of lines @xmath497 . the lines @xmath492 , @xmath498 share a point @xmath499 , the lines @xmath493 , @xmath500 share a point @xmath501 , and the lines @xmath498 , @xmath500 share a point @xmath502 , all four points @xmath503 being distinct .
next , pick some linear forms @xmath504 and @xmath505 representing @xmath498 and @xmath500
. then one can find constants @xmath506 so that the polynomial @xmath507 .
the argument is very similar to the one used in the proof of lemma 1 .
thus , @xmath508 , which can be written in the form @xmath509 required by @xmath469 . here , evidently , @xmath504 is not proportional to @xmath510 and @xmath505 is not proportional to @xmath32 .
therefore , for any @xmath372 that defines on @xmath93 an effective divisor containing a point of multiplicity two , @xmath511 must belong to the locus @xmath489 .
it follows `` by continuity '' that all the @xmath245 s that produce multiplicity functions @xmath360 , distinct from the identity function @xmath257 , project to @xmath489 via @xmath512 . according to lemma 5 ,
these are exactly the factorable polynomials that project to the `` combinatorial '' ramification locus @xmath513 .
@xmath107 it follows from lemma 6 that , for a generic @xmath514 , the affine subspace @xmath515 hits transversally the subvariety @xmath516 at a finite set of points whose cardinality is exactly @xmath332 .
therefore , @xmath332 must be the degree of that variety .
let @xmath365 be quasi - affine varieties and let @xmath517 be a proper regular map . in theorem 7 below
, the ramification set @xmath518 of a mapping @xmath517 with finite fibers is understood to be the closure of a set @xmath519 . by definition ,
@xmath520 when @xmath521 contains a point @xmath19 such that there is a non - zero vector @xmath522 that is mapped to zero in @xmath523 by the differential @xmath524 .
the arguments above lead to our main result : * the map @xmath525 is onto , and its generic fiber is a finite set of cardinality @xmath433 .
this map takes the multipole space @xmath297the space of a @xmath283-bundle @xmath526 over the variety @xmath527onto the complex vector space @xmath327 of dimension @xmath104 with its origin being deleted .
the map @xmath319 extends to a map @xmath528 with finite fibers .
* @xmath319 is ramified over the discriminant variety @xmath529the set of polynomials @xmath530 whose zero sets are effective divisors on the surface @xmath121 with at least one of their line components being of multiplicity at least two .
for which @xmath531 has a point of multiplicity at least two . ]
the subvariety @xmath532 is of complex codimension one , and is described in terms of solvability of @xmath391 or @xmath469 .
it contains the @xmath319-image of the singular set @xmath416 as a codimension one subvariety .
* in fact , @xmath319 , @xmath331 are @xmath283-equivariant maps . therefore , @xmath319 gives rise to a surjective map @xmath533 of degree @xmath428 which is also ramified over a subvariety @xmath534 of codimension one .
the compliment @xmath434 is a @xmath422-space , @xmath435 is a @xmath436-space , where @xmath423 is the braid group with coupling , while @xmath437 is a @xmath13-space , where @xmath14 an extension of @xmath438 by @xmath423 . *
the degree of the variety of completely factorizable homogeneous polynomials @xmath310 is also @xmath433 .
this variety is invariant under the obvious @xmath283-action on @xmath111 . @xmath107
theorem 7 has a number of topological implications , some of them dealing with interesting ramifications over complex projective spaces .
our next goal is to describe these implications .
let the space @xmath286 be of a homotopy type of a connected , finite - dimensional cw - complex .
recall that the dold - thom theorem [ dt ] links the homotopy groups of @xmath395 , @xmath1 being large , with the integral homology of a space @xmath286 . by picking a base point @xmath535 in @xmath286 ,
one gets a stabilization map @xmath536 , well - defined by the formula @xmath537 . here
this provides us with canonical homomorphisms @xmath539 of the @xmath275-th homotopy groups . in our case ,
the dold - thom theorem claims that @xmath540 .
in particular , @xmath541 . also , @xmath542 , provided @xmath543 or @xmath544 . on the other hand , @xmath545 , but @xmath546 . therefore , at least for large @xmath1 s , there is an infinite order element @xmath547 that is mapped by @xmath548 to zero and an element @xmath549 , so that @xmath550 is a generator .
it is possible to realize @xmath382 and @xmath551 geometrically .
what is clear ratherway , that the spheroids @xmath382 and @xmath551 can not be pushed into the aspherical portion @xmath435 of @xmath527 .
the realization of @xmath382 is based on an interesting fact that i learned from blaine lawson : the quotient of @xmath241 by the complex conjugation @xmath552 \rightarrow [ \bar x : \bar y : \bar z]$ ] is homeomorphic to the sphere @xmath553 .
therefore , the map @xmath554 given by the formula @xmath555 , where @xmath556 , is evidently @xmath557-equivariant with respect to the @xmath558-action in the domain and the symmetrizing action in the range .
this gives rise to the desired quotient map @xmath559 that survives into the higher symmetric powers of @xmath241 .
constructing class @xmath551 is straightforward : it is given by the obvious inclusion @xmath560 followed by the diagonal embedding @xmath561 .
the existence of a non - trivial @xmath562 whose @xmath403-image is null - homotopic in @xmath563 has a curious implication : for any @xmath564 , there is a family of polynomial functions of degree @xmath1 on the quadratic surface @xmath318 that is parameterized by a @xmath565-dimensional disk which does not admit a continuous lifting to the multipole space @xmath297 . at the same time , the functions parameterized by the @xmath566-sphere forming the boundary of the disk can be continuously represented by multipoles .
any map @xmath366 induces a natural map @xmath567 that is defined by the formula
@xmath568 = \sum_\nu \mu_\nu f(x_\nu)$ ] , where @xmath569 and @xmath570 . when @xmath37 is @xmath257-to-@xmath257 or onto , so is @xmath571 .
the construction of @xmath571 provides us with a rich source of interesting ramified coverings .
consider for example , a semi - free cyclic action on a sphere @xmath572 .
the group @xmath573 acts on @xmath2 by rotations around a fixed axis on the angles that are multiples of @xmath574 .
topologically , the orbit - space @xmath575 is again a 2-sphere .
let @xmath576 be the orbit - map .
then @xmath577 gives an example of a ramified degree @xmath578 covering map of @xmath579 over itself ! for @xmath580 , we shall see later how this degree @xmath581 ramification @xmath582 is linked to the multipole spaces on quadratic surfaces . on the other hand , by a simple cohomological argument
, any map @xmath583 has a degree that is the @xmath275-th power of a non - negative integer @xmath584 .
for instance , there is no ramified map from @xmath241 to itself of degrees that are not of the form @xmath585 .
* question * given a closed oriented manifold @xmath586 , what are possible degrees of maps from @xmath586 to itself ? evidently , the answer depends on the homotopy type of @xmath586 .
when @xmath557 acts freely on @xmath2 by the central symmetry , the orbit - map @xmath587 gives rise to another interesting ramified map @xmath588 of degree @xmath581 ( its existence follows from our results in section 3 dealing with real multipole spaces ) . in order to derive next few corollaries of theorem 7
, we need to take a detour aimed at constructing ( with the help of @xmath37 ) transfer maps that take effective @xmath589-divisors on @xmath590 to effective @xmath589-divisors on @xmath286 .
these constructions are variations on the theme of the classical hurwitz theorem ( see [ h ] , pp .
299 - 304 ) .
our next goal is to present constructions and arguments that lead to theorem 10 .
let @xmath286 and @xmath590 be smooth complex projective varieties and @xmath366 a regular surjective map with finite fibers . with any @xmath591 we associate a multiplicity number @xmath592 .
it is the multiplicity attached to the intersection of the @xmath37-graph @xmath593 with the subspace @xmath594 at the point @xmath595 . since all the @xmath37-fibers are finite , the intersection @xmath596 is finite as well .
next , with any @xmath597 we associate an effective @xmath589-divisor @xmath598 whose degree is @xmath20-independent and coincides with the degree @xmath599 of the map @xmath37 .
the correspondence @xmath600 produces a regular embedding @xmath601 for any @xmath275 , the map @xmath602 gives rise to an embedding @xmath603 defined by the formula @xmath604 .
therefore , applying this construction to the setting of theorem 7 with @xmath605 , @xmath606 , and @xmath607 , we get the following proposition : any irreducible quadratic form @xmath12 determines canonical embeddings @xmath608 where @xmath1 and @xmath275 are arbitrary whole numbers
. in particular , @xmath609 embeds @xmath610 into @xmath611 @xmath107 for a map @xmath366 as above , one can define a map @xmath612 that takes effective @xmath589-divisors on @xmath286 into effective @xmath589-divisors on @xmath286 . by definition , each point @xmath591 is mapped to the divisor @xmath613 . by linearity , we have @xmath614 this map transforms @xmath589-divisors of degree @xmath275 into @xmath589-divisors of degree @xmath615 : @xmath616 both maps , @xmath617 from @xmath34 and @xmath618 from @xmath619 , have the same targets
. moreover , their images coincide .
indeed , for each point @xmath597 , pick any point @xmath620 .
then , @xmath621 .
recall , that @xmath618 is a 1-to-1 map and thus is invertible over its image .
therefore , the map @xmath622 is well - defined .
let @xmath286 and @xmath590 be smooth complex projective varieties and @xmath366 be a regular onto map with finite fibers .
the map @xmath623 in @xmath624 coincides with the natural map @xmath567 . it defines a ramified covering with a generic fiber of cardinality @xmath625 .
* proof * the map @xmath612 takes each point @xmath591 to the divisor @xmath613 . in @xmath613 is @xmath592 . ] . at the same time
, the transfer @xmath602 takes @xmath626 to the same divisor @xmath613 .
thus , @xmath627 maps @xmath19 to @xmath626 . extending this argument by linearity
proves the claim .
@xmath107 now we examine how these constructions apply to projective curves in @xmath52 and eventually to the multipole spaces .
the role of @xmath286 will be played by a curve @xmath628 ( most importantly , by @xmath93 ) , the role of @xmath590 by a pencil of lines in @xmath52 through a point @xmath94 .
the map @xmath37 takes any point @xmath629 to a line @xmath244 passing through @xmath266 and @xmath94 .
any linear embedding @xmath630 induces an embedding @xmath631 .
the geometry of @xmath632 is tricky .
since the quotient space @xmath633 is homeomorphic to a 4-sphere @xmath553 , we get a homeomorphism @xmath634 but even the spaces @xmath635 have subtle topology .
for example , @xmath636 is homeomorphic to a mapping cylinder of a map @xmath637 , where @xmath638 denotes the fourth suspension ( see [ ha ] , example 4k.5 ) .
the regular map @xmath639 describes the role and place of @xmath46-_planar _ multipoles ( they are linked to the planarity of the quadra- poles and octapoles in the deconstruction of cmbr that was briefly mentioned in the introduction ) .
in contrast with @xmath403 , we will see that the map @xmath640 is 1-to-1 .
since @xmath641 is homeomorphic to @xmath642 , its image in @xmath408 is homeomorphic to @xmath643 as well ( compare this with @xmath644 ) .
moreover , the @xmath59-cycle @xmath645 $ ] is homologous to the cycle @xmath646 $ ] ( equivalently , to the cycle @xmath647 $ ] ) defined by a natural imbedding @xmath648 ( equivalently , by a linear imbedding @xmath649 ) .
recall that points @xmath650 correspond to a pencil of lines @xmath651 that pass through a particular point @xmath652 .
that @xmath94 determines the embedding @xmath653 .
each point @xmath499 determines a unique line @xmath654 that passes through @xmath266 and @xmath94 , and therefore , a unique point @xmath655 in the dual subspace @xmath656 .
let @xmath657 be a @xmath15-to-@xmath257 ramified map defined by the correspondence @xmath658 .
as described in ( 24 ) and ( 26 ) , @xmath37 gives rise to maps @xmath659 and @xmath660 with @xmath661 being a @xmath257-to-@xmath257 map .
in fact , examining the construction of @xmath403 , we see that @xmath662 . moreover , using lemma 7 , the map @xmath663 factors as @xmath664 . therefore , the ramification locus @xmath665 for @xmath666 is the @xmath667-image of the ramification locus @xmath668 for @xmath669 .
figure 2 illustrates the case @xmath670 . passing from the first to the second column depicts the map @xmath671 generated by the linear projection from a center @xmath94 located at infinity .
passing from the first to the third column depicts the map @xmath672 . here
@xmath386 is viewed as the pencil of lines in @xmath52 through @xmath94 .
topologically , the map @xmath673 is a 8-to-1 ramification of @xmath674 over itself .
it is described in some detail in example 1 and is depicted in figure 3 .
the passage from the second to the third column is a 1-to-1 correspondence .
when @xmath654 is not tangent to @xmath93 at @xmath266 , then let @xmath675 be `` the other point '' in @xmath93 that belongs to the line @xmath654 .
when @xmath654 is tangent to @xmath93 at @xmath266 , then by definition , put @xmath676 .
the correspondence @xmath677 is an involution on @xmath93 with two fixed points @xmath535 and @xmath678 .
its orbit - space @xmath679 topologically is a @xmath15-sphere .
the involution @xmath680 induces an involution @xmath681 that acts on the space @xmath682 . by definition ,
any effective divisor @xmath683 is transformed by @xmath681 into the divisor @xmath684 .
evidently , the image of @xmath660 is contained in the @xmath685-invariant part of the space @xmath408 ; however , not any invariant divisor belongs to that image .
each divisor from @xmath686 not only must be @xmath687-invariant , but in addition , its multiplicity at the fixed points @xmath535 and @xmath678 must be _
even_. in other words , such a divisor @xmath688 must be of the form @xmath689,\ ] ] where @xmath690 and @xmath691 .
the cardinality of the @xmath692-fiber over @xmath688 is given by the formula @xmath693 there are @xmath694 effective divisors of degree @xmath695 with the support in @xmath696 and whose @xmath697-image is @xmath698 $ ] ; at the same time , there is a unique divisor @xmath699 whose @xmath700-image is @xmath701 .
when all @xmath702 and @xmath703 , the fiber @xmath704 is of cardinality @xmath705 .
now , we are in position to justify previous claims about the image of the map @xmath640 in ( 28 ) . pick a divisor @xmath706 of degree @xmath1 in @xmath93 such that the supports of @xmath706 and @xmath707 do not share common points . for any divisor @xmath108 of degree @xmath1 on @xmath93
the operation @xmath708 gives rise to an imbedding @xmath709 .
we claim that the intersection @xmath710 is a singleton .
indeed , @xmath711 consists of @xmath687-invariant divisors , but the only invariant divisor of the form @xmath712 is the divisor @xmath713 .
therefore , using the cohomology ring structure of @xmath714 , the cycle @xmath715 \in h_{2d}(sym^{2d}(\mathcal q ) ; \mathbb z)$ ] must be equal to the cycle @xmath716 $ ] . using the diffeomorphism @xmath717 ,
this translates into the generating cycle @xmath647 \in h_{2d}(\mathbb cp^{2d } ; \mathbb z)$ ] .
examining @xmath718 , we see that the ramification set @xmath719 for @xmath720 is comprised of divisors of two kinds : 1 ) the ones that contain at least one summand of the form @xmath721 , where @xmath722 , and 2 ) the ones that contain @xmath723 or @xmath724 as a summand . for example , if @xmath725 contains a pair of distinct points @xmath726 and the rest of points in the support of @xmath108 are generic ( i.e. , for @xmath727 , @xmath728 ) , then @xmath729 consists of @xmath730 elements . at the same time , as we perturb @xmath108 in order to avoid the coincidence @xmath731 , @xmath729 consists of @xmath705 elements . therefore , in view of lemma 7 , the ramification set @xmath732 for @xmath663 is comprised of divisors of two kinds : 1 ) the ones that contain at least one summand of multiplicity at least two , and 2 ) the ones that contain points @xmath733 or @xmath734 giving rise to lines @xmath735 and @xmath736 passing through the point @xmath94 and tangent to the curve @xmath93 . given a space @xmath286 ,
let us denote by @xmath737 the discriminat set formed by the divisors containing points of multiplicity at least two .
also , for any point @xmath738 , we denote by @xmath739 the subset of @xmath395 formed by the divisors containing the summand @xmath740 .
thus , @xmath741 and @xmath742 employing the vite map @xmath743 , we transplant the algebraic set @xmath732 into the space @xmath642 .
the image of @xmath744 under the vite map is the classical discriminant variety in @xmath745 , while the images of @xmath746 and of @xmath747 form linear subspaces @xmath748 and @xmath749 in @xmath642 .
thus , @xmath750 .
it follows from [ k ] , theorem 6.1 , that each of the two spaces @xmath748 and @xmath749 is _ tangent _ to the discriminant variety @xmath751 along , respectively , the linear subspaces @xmath752 and @xmath753 .
note that the complement @xmath754 is the configuration space of @xmath1-tuples of distinct points in the domain @xmath755 . here
@xmath756 stand for the two points in @xmath386 that are dual to the lines passing through @xmath94 and tangent to the curve @xmath93 .
therefore , it is a @xmath757-space , where @xmath758 stands for the braid group in @xmath1 strings residing in a cylinder with an annulus base @xmath759 \times [ 0,1]$ ] . using the birational identifications
@xmath760 , we have constructed a ramified covering @xmath761 .
the following proposition summarizes the conclusions of our arguments above ( centered on @xmath762 and @xmath763 ) .
it describes an intricate stratified geometry of this ramified covering , a geometry that is reductive " in its nature with respect to the shift @xmath764 . * any irreducible quadratic form @xmath66 gives
rise to a ramified covering @xmath765 of degree @xmath705 .
the map @xmath766 is ramified over the algebraic set @xmath767 the vite image of the set @xmath768 * the discriminant variety @xmath769 is @xmath66-independent .
the two linear subspaces @xmath748 and @xmath749 are tangent to the variety @xmath751 along , respectively , subspaces @xmath752 and @xmath753 . a generic point of @xmath751 has a @xmath766-fiber of cardinality @xmath770 , less than a generic @xmath766-fiber . ] while a generic point of @xmath771 has a fiber of cardinality @xmath772 . *
the complement @xmath773 to the ramification set is a @xmath757-space , where @xmath758 denotes the annulus braid group in @xmath1 strings .
* moreover , over to each of the two subspaces @xmath774 and @xmath775 , the map @xmath766 inherits a similar stratified structure with respect to the dimensional shift @xmath764 .
@xmath107 and the geometry of its ramification locus.,width=480,height=384 ] * example 1 .
* let @xmath17 .
then @xmath776 is of degree @xmath566 .
the discriminat parabola @xmath777 is given by @xmath778 .
the ramification locus @xmath779 consists of that parabola together with two tangent lines @xmath780 and @xmath781 which share a point @xmath782 $ ] .
they are tangent to the parabola at the points @xmath783 $ ] and @xmath784 $ ] . here
the parameters @xmath785 and @xmath786 depend on the quadratic form @xmath12 .
the @xmath766-fibers over @xmath787 are singletons ; the cardinality of the fiber over @xmath788 is @xmath18 ; the cardinality of the fiber over @xmath789 and over @xmath790 is @xmath15 .
it is still a bit mysterious how all this data manage to produce @xmath241 as a covering space !
anyway , the compliment @xmath791 is a @xmath792-space , where @xmath793 is the braid group with two strings in a `` fat '' annulus . now consider the case @xmath670 depicted in figure 3 .
that figure exhibits a two - fold symmetry that exchanges points @xmath535 and @xmath678 where the pencil of parallel lines is tangent to the curve @xmath93 .
pattern 1 in figure 3 corresponds to the generic stratum @xmath674 , while pattern 2 defines the discriminant surface @xmath794 of degree @xmath566 .
it is homeomorphic to @xmath795 .
patterns 3a and 3b each corresponds to two planes @xmath796 and @xmath797 .
each of the two planes is tangent to the discriminant surface @xmath798 along lines @xmath799 and @xmath800 labeled by patterns 6a and 6b , respectively .
these planes intersect along another line @xmath584 labeled by pattern 5 .
pattern 4 encodes the singular locus @xmath801 of the discriminant surface @xmath798 .
it is a rational curve of degree 3 that is homeomorphic to @xmath2 .
in fact , @xmath798 is the ruled surface spanned by lines tangent to @xmath801 ( [ k ] ) .
the surface @xmath798 intersects with the plane @xmath796 along the union of the line @xmath799 with a quadratic curve @xmath802 labeled by pattern 7a .
the lines @xmath799 and @xmath584 are both tangent to the parabola @xmath802 .
this configuration is already familiar from our description of the case @xmath17 .
similarly , pattern 7b labels parabola @xmath803 .
curves @xmath801 , @xmath802 , and @xmath799 are all tangent at a point @xmath804 labeled by pattern 8a .
curves @xmath584 and @xmath802 are tangent at a point @xmath805 labeled by pattern 9b .
that point also lies on the line @xmath800 .
the labeling of points @xmath806 and @xmath805 is done by patterns 8b and 9a , respectively . in accordance with theorem 10 , the ramification locus @xmath779 for the 8-to-1 map @xmath766 coinsides with @xmath807 and its complement is a @xmath808-space .
@xmath107 now , we turn to deconstructions of both homogeneous non - homogeneous polynomials on complex quadratic surfaces @xmath809 .
first , consider a general homogeneous polynomial @xmath8 of degree @xmath1 .
we can apply decomposition ( 6 ) to the homogeneous term @xmath24 of degree @xmath25 in ( 6 ) .
this will produce a new collection of vectors a new multipole@xmath810 associated with an appropriate generalized parcelling of @xmath811 . this process of producing lower order multipoles @xmath812 , where @xmath813 and @xmath814 , can be repeated again and again until all the degrees are used up " .
we notice that the leading multipole ( of highest degree ) is determined by this algorithm in a more direct way " than the lower degree multipoles .
also note that the choice of a generalized parcelling @xmath558 for @xmath249 affects the choice of a generalized parcelling for @xmath815 , where the polynomial @xmath816 . here
the product @xmath97 is determined by the @xmath558 , and @xmath817 is the maximal power of @xmath66 for which the division in the ring of polynomials is possible .
* example 2 .
* let @xmath818 .
then any homogeneous polynomial @xmath8 of degree @xmath565 has a representation of the form @xmath819 where all the @xmath820 s are linear forms ( some of which might be zeros ) .
the number of such representations does not exceed @xmath821 .
@xmath107 any non - homogeneous polynomial @xmath0 of degree @xmath1 can be written in the form @xmath822 where the degrees of monomials comprising @xmath823 are congruent to @xmath225 modulo @xmath15 .
note that the decomposition @xmath824 intrinsically makes sense on every quadratic surface @xmath825 : if @xmath8 belongs to the principle ideal @xmath826 , so does each term @xmath827 .
thus , if @xmath828 , then we have @xmath829 . on the surface @xmath830 ,
any component @xmath823 of the polynomial @xmath822 can be homogenized by multiplying its terms of the same degree by an appropriate power of @xmath66 .
we denote by @xmath831 the appropriate homogeneous polynomial .
generically , @xmath832 .
* example 3 .
* let @xmath818 .
any polynomial @xmath8 of degree @xmath565 has a representation of the form @xmath833 where all the @xmath820 s and @xmath834 s are linear forms ( some of which might be zeros ) , and @xmath99 is a number .
the number of such representations does not exceed @xmath835 .
@xmath107 now one can apply recursively lemmas 1 , 3 and corollary 5 to each homogeneous polynomial @xmath831 , @xmath836 .
letting @xmath837 proves formula @xmath43 from theorem 2 .
let us restate and generalize this theorem in terms of the multipoles : * let @xmath12 be an irreducible quadratic form and let @xmath0 be any complex polynomial of degree @xmath1 .
its restriction @xmath838 to the complex surface @xmath839 admits a representation of the form @xmath840 where the linear forms @xmath841 are chosen so that each non - zero product @xmath842 is determined , via the map @xmath319 ( see @xmath396 ) , by an appropriate multipole @xmath843 from the variety @xmath274 . *
the representation @xmath844 is unique , up to a finite ambiguity and up to reordering and rescaling of multipliers in each product @xmath842 . in other words ,
the set of @xmath8-representing multipoles @xmath845 is finite .
its cardinality does not exceed @xmath846 $ ] .
@xmath107 we would like to end this section by establishing a few facts about alternative decompositions of polynomials that are based on formula @xmath847 . in achieving this goal
we are guided by theorem 22.2 from [ sh ] .
we claimed that the direct sum in @xmath132 is _ orthogonal _ with respect to the inner product @xmath848 , where @xmath134 is the standard rotationally symmetric measure on the unit sphere @xmath2 .
let us clarify this claim .
applying @xmath132 recursively , we get that for any real homogeneous polynomial @xmath8 of degree @xmath1 can be written as @xmath849 , where @xmath850 is a real homogeneous harmonic polynomial of degree @xmath851 . letting @xmath837
, we see that for any homogeneous @xmath8 , there is a harmonic polynomial @xmath852 such that @xmath853 .
now , any two harmonic homogeneous polynomials @xmath850 and @xmath854 of different degrees are eigenfunctions with different eigenvalues @xmath855 and @xmath856 for the laplace operator @xmath10 on the sphere . therefore , they are orthogonal , i.e. @xmath857 . since any homogeneous polynomial @xmath24 of degree @xmath25 can be represented as @xmath858 , we get the claimed orthogonality @xmath859 of the direct sum in @xmath132 .
this argument extends to complex harmonic polynomials as follows .
we already remarked that the real and imaginary parts of a complex harmonic polynomial @xmath8 are real harmonic polynomials of the same degree . if the polynomial is homogeneous , so are its real and imaginary parts .
therefore , for any complex harmonic and homogeneous polynomial @xmath8 of degree @xmath1 and any homogeneous polynomial @xmath24 of degree @xmath25 , we get @xmath860 \ ; dm = \int_{s^2 } ( \mathsf{re } p + \mathbf i\ ; \mathsf{im } p ) ( \mathsf{re } r - \mathbf i\ ; \mathsf{im } r ) \ ; dm = \int_{s^2 } ( \mathsf{re } p \cdot \mathsf{re } r ) \ ; dm + \int_{s^2 } ( \mathsf{im } p \cdot \mathsf{im } r)\ ; dm - \mathbf i \int_{s^2 } ( \mathsf{re } p \cdot \mathsf{im } r)\ ; dm + \mathbf i \int_{s^2 } ( \mathsf{im } p \cdot \mathsf{re } r)\ ; dm$ ] . each of the four integrals must vanish because each integrant is a product of a real homogeneous and harmonic polynomial of degree @xmath1 by a real polynomial of a lower degree
. in the spherical coordinates @xmath861 , the inner product is given by an integral @xmath862 where @xmath863 .
now we are going to transfer this hermitian inner product from the sphere to its image @xmath864 in a given complex quadratic surface @xmath865 .
as before , let @xmath785 be a complex invertible matrix that reduces @xmath66 to the sum of squares : @xmath866 , where @xmath867 .
let @xmath868 , with @xmath869 being a _ real _ vector .
the ellipsoid @xmath864 is the image of the unit sphere under the complex linear transformation @xmath870 .
it is a totally real 2-dimensional algebraic surface in the complex surface @xmath871 .
the real - valued measure @xmath872 on @xmath864 is the pull - back under the map @xmath785 of the standard measure on the unit sphere .
it is invariant under the action of the compact subgroup @xmath873 , where @xmath874 denotes the orthogonal group . for any pair of ( homogeneous ) complex functions @xmath875 on @xmath157 , we get @xmath876 applying @xmath847 recursively , gives the following proposition : * the space of complex homogeneous polynomials admits an @xmath192-invariant decomposition @xmath877 the summands in @xmath878 are orthogonal with respect to the hermitian inner product defined by the formula @xmath879 . in particular , the homogeneous polynomials @xmath880the solutions of the equation @xmath218are characterized by the property : for each @xmath881
, @xmath882 * any polynomial function @xmath883 on the surface @xmath884 can be obtained by restricting to @xmath53 a polynomial @xmath215 .
is homogeneous ] * for any two polynomials @xmath586 and @xmath885 , the complex dirichlet problem " @xmath886 has a unique _ polynomial _
solution @xmath8 . * any homogeneous polynomial @xmath8 of degree @xmath1 admits a maxwell - type representation @xmath887 * remark * all the statements of theorem 12 , but the last one ( dealing with the generalized maxwell representation @xmath888 ) , can be easily generalized for polynomials in any number of variables . * proof * decomposition @xmath878 follows from @xmath847 , and @xmath888 from @xmath878 together with @xmath213 .
the second bullet is obviously implied by @xmath878 and the linearity of @xmath11 . in order to prove the third bullet , note that @xmath847 implies that @xmath889 is onto .
therefore , for any @xmath586 , there exists a polynomial @xmath158 , so that @xmath890 . on the other hand by bullet two , there exists a harmonic polynomial @xmath891 such that @xmath892 . as a result , @xmath893 solves the dirichlet problem . in order to show that @xmath8 is unique , it is sufficient to prove that no polynomial of the form @xmath894 belongs to @xmath200 . since @xmath11 is homogeneous of degree @xmath895 , @xmath896 , unless @xmath897 .
indeed , let @xmath883 be the leading homogeneous portion of @xmath898 of the degree @xmath899
. then @xmath900 implies @xmath901 which is , as we have shown before , impossible , unless @xmath902 .
we have seen that different parcellings of the intersection @xmath56 led to different deconstructions of polynomial functions on a quadratic surface @xmath903 .
the next idea is to invoke symmetry in order to get a unique equivariant parcelling and thus , a _
unique _ deconstruction of symmetric functions on a quadratic surface supporting an appropriate group action .
we deal with a special , but important case of @xmath557-action .
the model example is provided by the complex conjugation @xmath904 . if @xmath12 has real coefficients , both the curve @xmath905 and the surface @xmath906 are invariant under the @xmath558 .
when @xmath66 is a definite real form , the @xmath557-action by conjugation on @xmath93 is _ free _ since the fixed point set @xmath907 . for a homogeneous polynomial @xmath8 with real coefficients
the intersection @xmath908 and the multiplicity function @xmath249 are @xmath558-invariant as well .
we notice that a @xmath558-invariant generalized parcelling of such a @xmath67 will produce a set of @xmath558-invariant lines @xmath96 .
thus , we can assume that all the linear forms @xmath88 in @xmath130 are with real coefficients .
indeed , given a representation @xmath909 , where @xmath910 and @xmath88 s are real and @xmath47 , we get @xmath911 . hence , @xmath912 which implies the validity of @xmath130 over the reals . with this observation in mind , we introduce real versions of the multipole spaces ( 11 ) , ( 12 ) .
let @xmath913^k / \sigma_k^\mathbb r\end{aligned}\ ] ] where the group @xmath914 is defined similar to its complex version .
the only difference lies in the definition of @xmath915 : it is now a subgroup of @xmath916 , not of @xmath917 .
the space of real multipoles @xmath918 is a space of a principle @xmath919-fibration over the real variety @xmath920 .
we can form an associated line bundle @xmath921 .
as in the complex case , put @xmath922 and form the quotient space @xmath923 by collapsing the zero section to a point @xmath290 .
as before , @xmath924 is a contractible space .
real homogeneous polynomials of degree @xmath1 form a totally real subspace @xmath925 in the hermitian space @xmath111 .
the hermitian metric on @xmath111 defined by @xmath879 generates an eucledian metric on @xmath925 induced by the inner product @xmath926 moreover , as in the complex case , the imbedding @xmath927 is an isometry ( recall that @xmath66 is real ) .
similarly to the complex case , we introduce a variety @xmath928 of completely factorable real polynomials .
the orthogonal projection @xmath929 maps @xmath930 into the vector space @xmath931 .
due to theorem 12 , @xmath931 can be identified with @xmath932 . when the quadratic form @xmath66 is not definite
, we can also identify the space @xmath931 with the @xmath1-graded portion of the polynomial function ring on the real cone @xmath318 . in any case , we get an algebraic map @xmath933 which , by lemma 1 and the argument above , is _
onto_. as in the complex case , this map extends to a map @xmath934 that takes the zero section @xmath935 of the line bundle @xmath936 to the origin in @xmath931 , and each fiber of @xmath936 isomorphically to a line through the origin .
this generates an @xmath919-equivariant surjection @xmath937 with finite fibers , where @xmath938 .
the number of @xmath558-equivariant parcellings of @xmath56 is harder to determine .
it depends only on the restriction of the multiplicity function @xmath67 to the subset @xmath939 and is equal to the number of generalized parcellings subordinate to such a restriction . however , if @xmath940 , the conjugation acts freely on @xmath905 and the @xmath558-equivariant parcelling is unique .
in such a case , we are getting a more satisfying result : let @xmath12 be an irreducible quadratic form with real coefficients and the signature distinct from @xmath941 .
then any real polynomial @xmath0 of degree @xmath1 , being restricted to the real conic @xmath942 in @xmath943 , has a representation @xmath944,\end{aligned}\ ] ] where each @xmath945 is a unit vector and , for @xmath946 , @xmath947 .
when the surface @xmath948 is an ellipsoid , the represention @xmath949 is unique , up to reordering and rescaling of the multiplyers in the products . in such a case , @xmath8 gives rise to a unique sequence of multiploles @xmath950 .
@xmath107 we observe that @xmath951 happens to be a nonsingular space : locally @xmath952 and @xmath386 are diffeomorphic , and @xmath953 is non - singular .
an @xmath954-version of the argument that follows theorem 6 and is centered on formula @xmath469 is valid .
therefore , an @xmath954-version of lemma 6 holds : the combinatorial " and the smooth " ramification loci coincide . note that one always can find a real homogeneous polynomial @xmath8 of degree @xmath275 for which the curves @xmath955 have an empty intersection : just take a linear form @xmath245 such that the line @xmath956 misses the real quadratic curve @xmath93 ; then consider any polynomial @xmath8 sufficiently close to @xmath957 .
for such a @xmath8 , the @xmath557-equivariant parcelling is unique and , thus , @xmath958 is a singleton .
evidently , such polynomials @xmath8 form an open set . therefore , @xmath959 is a map of degree one .
these remarks lead to the multipole space @xmath951 , which parametrizes completely factorable real homogeneous polynomials of degree @xmath275 , is a space of an @xmath919-bundle over the real variety @xmath960 .
the @xmath919-equivariant map @xmath961 induces a surjective mapping @xmath962 of degree one . unless @xmath66 is a definite form , the map @xmath961 is ramified over the discriminant variety @xmath963 of codimension one . here
@xmath963 is comprised of homogeneous polynomials @xmath8 of degree @xmath275 on the real surface @xmath318 for which the surfaces @xmath964 and @xmath318 share a line of multiplicity at least two . and
@xmath93 in @xmath965 share a point of multiplicity at least two . ]
@xmath107 let @xmath66 be a positive definite form .
then * the multipole space @xmath966^d \big\}\big/\sigma_d^\mathbb r$ ] is diffeomorphic to the space @xmath967 . *
the multipole space @xmath968 in @xmath969 is homeomorphic to the space @xmath970 . * the real varieties @xmath971 and @xmath972 are diffeomorphic . * proof * in this case , the ramification locus @xmath973 .
since under the corollay s hypotheses the @xmath558-equivariant parcelling is unique , every real homogeneous @xmath8 , not divisible by @xmath66 , gives rise to a unique leading multipole @xmath974the map @xmath975 is @xmath257-to-@xmath257 .
recalling that @xmath976 is onto a vector space of dimension @xmath104 and that the smooth ramification locus @xmath977 , completes the proof .
@xmath107 the diffeomorphism @xmath978 serves as yet another transparent illustration to the dold - thom theorem : not only it reveals @xmath979 as a @xmath980 space , but we actually know how this stabilization of the homotopy groups occurs . in fact , @xmath981 , and for @xmath982 , @xmath983 .
in particular , @xmath984 vanish for @xmath985 .
we have seen already the main advantages of the multipole representations for the polynomials on quadratic surfaces : such representations are independent on the choice of coordinates in @xmath49 or @xmath943 .
this is in the sharp contrast with the classical decompositions in terms of the spherical harmonics . in the case of @xmath81 and over the reals ,
the independence of the multipoles under the rotations was observed by many . when @xmath66 is not positive - definite , similar observations hold . for a non - degenerated quadratic form @xmath66 with real coefficients , let @xmath986 denote the group of linear transformations from @xmath987 that preserve the form .
when the signature @xmath988 , then @xmath986 contains the lorenz transformation group ( equivalently , the isometry group of a hyperbolic plane ) as a subgroup of index two .
the next proposition should be compared with lemma 2 and theorem 12 .
we noticed already that , for a _ real _ quadratic form @xmath66 , the decomposition ( 9 ) is a complexification of a similar decomposition @xmath989 over the reals . here
therefore , ( 46 ) is @xmath986-equivariant with respect to the natural action on the space @xmath129 and orthogonal with respect to the inner product @xmath991 . at the same time
, the multipole space @xmath992 is also equipped with the @xmath986-action induced by the obvious diagonal action on @xmath993^d$ ] .
furthermore , the map @xmath994 induced by the projection @xmath995 defined by ( 46 ) is equivariant . given a real polynomial @xmath8 on @xmath948 together with its representation @xmath949 and a transformation @xmath996 , the transformed polynomial @xmath997 on @xmath948 acquires a representation in the form @xmath998,\end{aligned}\ ] ] where each new multipole vector @xmath999 .
in other words , the onto map @xmath1000 \ ; \approx \oplus_{k = 0}^d \ ; har_q(d ; \mathbb r)\ ] ] is @xmath986-equivariant .
@xmath107 combining decomposition @xmath1001 with theorem 12 we get its real analog : let @xmath66 be a real non - degenerated quadratic form .
* the space of real homogeneous polynomials admits an @xmath1002-invariant decomposition @xmath1003 the summands in @xmath1004 are orthogonal with respect to the inner product defined by the formula @xmath991 . * any polynomial function @xmath883 on the surface @xmath1005
can be obtained by restricting to @xmath1006 a polynomial @xmath1007 .
* for any two polynomials @xmath586 and @xmath885 , the dirichlet problem @xmath1008 has a unique real _ polynomial _
solution @xmath8 . * any real homogeneous polynomial @xmath8 of degree @xmath1 admits the maxwell - type representation @xmath1009 where @xmath1010 are complex 3-vectors .
these vectors are real and representation @xmath1011 is unique , provided that @xmath66 is positive - definite .
let @xmath12 be an irreducible form of degree @xmath584 over @xmath46 .
then @xmath1012 is the surface in @xmath49 and @xmath1013 is an irreducible curve in @xmath52 of degree @xmath584 . as before ,
we denote by @xmath115 the set of homogeneous degree @xmath1 complex polynomials that are divisible by @xmath66 .
let @xmath1014 be an orthogonal complement to @xmath115 in @xmath111 . for any sequence of non - negative integers
@xmath1015 so that @xmath1016 , consider a map @xmath1017 which is defined by taking the product of homogeneous polynomials @xmath1018 restricted to the surface @xmath121 . as before , the subgroup @xmath1019 of rank @xmath1020 acts freely on @xmath1021 $ ] by scalar multiplication . by the definition of @xmath1022 ,
the map @xmath1023 is constant on the orbits of this action .
we view the partition @xmath1024 as a non - increasing function @xmath1025 on the index set @xmath1026 .
denote by @xmath1027 the subgroup of the permutation group @xmath1028 that preserves @xmath1029 .
let @xmath1030 be an extension of @xmath1027 by @xmath1022 that is generated by the obvious actions of @xmath1027 and @xmath1022 on @xmath1031 .
evidently , @xmath1023 is an @xmath1030-equivariant map .
thus , it gives rise to a well - defined map .
] @xmath1032 because @xmath66 is irreducible , the map @xmath1033 has @xmath1034 , and not just @xmath119 , as its target . as before , we introduce the mulipole space @xmath1035 which is a space of a principle @xmath283-fibration over the orbifold @xmath1036 as in the case of quadratic forms @xmath66 , one has a map @xmath1037 , where @xmath1038 is the variety of homogeneous degree @xmath1 polynomials in @xmath5 , and @xmath6 that admit a factorization as a product of polynomials of the degrees @xmath1039 prescribed by the partition @xmath1029 .
the map @xmath1040 takes each multipole to the product of the corresponding polynomial factors . unlike the case of a quadratic @xmath1041 @xmath1040
may not be a 1-to-1 map , although , its generic fiber is a singleton .
this conclusion follows from the unique factorization property for the ring @xmath44 $ ] : just consider elements of @xmath1042 that are products of _ irreducible _ factors of the degrees prescribed by @xmath1029 .
the same uniqueness of factorization implies that each fiber of @xmath1040 is finite : there are only finitely many ways of organizing irreducible factors , in which a polynomial @xmath1043 decomposes , into blocks of degrees @xmath1015 .
the map @xmath1040 is not surjective either .
however , @xmath1033 takes the multipole space _ onto _ the space @xmath1044 of degree @xmath1 homogeneous polynomials on the surface @xmath318 that admit factorizations
subordinate to @xmath1029 : @xmath1045 let @xmath1 be a natural number and @xmath1046 its partition .
let @xmath72 be a finite set equipped with a multiplicity function @xmath73 whose @xmath74-norm @xmath75 is @xmath1047 .
a _ generalized _ @xmath1029-_parcelling _ of @xmath76 is a collection of functions @xmath1048 , such that * @xmath1049 * @xmath1050 when @xmath72 is comprised of @xmath1047 points and each @xmath1051 takes only two values @xmath1052 , the generalized parcelling is called just an @xmath1029-_parcelling_. any polynomial @xmath0 of degree @xmath1 that is not divisible by @xmath66 determines a multiplicity function @xmath83 whose @xmath74-norm is @xmath1047 . here
@xmath1053 is a finite set . if such a polynomial @xmath8 is a product @xmath1054 , where @xmath1055 , then the @xmath1056 s give rise to a unique generalized @xmath1029-parcelling @xmath1057 .
any generalized @xmath1029-parceling @xmath1057 of a given multiplicity function @xmath67 on a finite set @xmath1058 corresponds to at most one multipole in the space @xmath1059 . * proof * assume that , for each index @xmath1060 , there exist a polynomial @xmath1056 that realizes @xmath1051 on the finite intersection set @xmath1058 .
such polynomial is not divisible by @xmath66 .
put @xmath1061 .
then , employing the bezout theorem , any other polynomial @xmath1062 that realizes the same multiplicity on the same intersection set @xmath1063 must be of the form @xmath1064 .
we notice that if @xmath1065 , then @xmath1066 , provided @xmath1067 .
@xmath107 the map @xmath1033 has finite fibers over @xmath1044 .
* proof * an element of @xmath1068 is determined , up to proportionality , by its multiplicity function @xmath343 .
now the claim of the corollary follows from lemma 8 and the observation that a given multiplicity function admits only finitely many generalized @xmath1029-parcellings .
@xmath107 of course , not any generalized @xmath1029-parceling on a given pair @xmath1069 is realizable by a product @xmath1054 with the properties as above .
crudly , this happens because not any @xmath1070 points on @xmath93 can be placed on a curve @xmath1071 of degree @xmath1072 that does not contain @xmath93 as its component .
a curve of degree @xmath1072 can always accommodate @xmath1073 points in @xmath52 .
thus , if an inequality @xmath1074 is valid , that is , if @xmath1075 , the right curve might be found ; but it is still unclear how to avoid the very real possibility that @xmath1076 when @xmath1077 .
in fact , such a possibility is a reality ! unfortunately , unless @xmath1078 or @xmath1079 , the image of @xmath1033 is of a smaller dimension than the one of the target space @xmath1034 . as a result ,
_ there is no analog of the sylvester theorem in the non - quadratic skies _ ; when @xmath1080 , a generic element of @xmath1034 is _
irreducible_. let us explain these claims .
recall that for any @xmath1081 , @xmath1082\ } = \frac{l}{2}(2d - l + 3)$ ] . since the dimension of the group @xmath1022 is @xmath1083 , we get @xmath1084,\ ] ] provided all @xmath1077 . under these hypotheses ,
the difference of the two dimensions vanishes only when @xmath1085 or @xmath1079 .
hence , for @xmath1086 and @xmath1087 , the map @xmath1033 is not onto , i.e. a generic polynomial from @xmath1034 is not a product of polynomials of degrees @xmath1088 .
the codimension of @xmath1089 in @xmath119 is @xmath1090 $ ] .
for example , on a cubic surface , @xmath1091 , provided @xmath1092 .
if we drop the hypotheses @xmath1093 , the computation is a bit more involved : @xmath1094
we conjecture that the rhs of the formula above is always positive , unless @xmath1095 or @xmath1096 .
we will be concerned with polynomial approximations of holomorphic functions @xmath1097 on an irreducible complex quadratic surface @xmath1098 , as well as with polynomial approximations of continuous functions @xmath1099 on its real version .
eventually , we would like to represent the approximating polynomials in terms of their multipoles . as one uses polynomials of higher and higher degrees to improve approximations , the issue is stability of the multipole representations . in general ,
such stability is absent for several reasons : 1 ) the intrinsic ambiguities of the multipole representations for complex plynomials ; 2 ) the impossibility of converting an analytic function on a surface into a
homogeneous " analytic function in the ambient space ( homogenizing polynomials worked well ) . even abandoning multipole representations in favor of linear methods of harmonic analysis ,
does not eliminate the stability issue instantly : in general , the coefficients of approximating polynomials fail to stabilize .
however , if the approximating polynomials are linear combinations of mutually _ orthogonal _ and normalized polynomials ( analogous to the legendre polynomials ) , the coefficients of the combinations will stabilize . by introducing appropriate notions of orthogonality for polynomials on a quadratic surface , we aim to establish similar facts for polynomial approximations there ( the spherical harmonics reflect a particular case of such orthogonality )
. then we can combine harmonic analysis with non - linear methods of multipole representation for polynomials .
we call such an approach _
synthetic_. let @xmath1100 denote the algebra of all continuous @xmath46-functions on a hausdorff compact space @xmath1101 .
recall that a _
uniform algebra _ is a closed ( in the @xmath38-norm ) subalgebra @xmath1102 that separates points of @xmath1101 .
such an algebra is called _ antisymmetric _ , if any real - valued function from @xmath1103 is constant .
a subset @xmath1104 is called an _
antisymmetry set _ for @xmath1103 if any function from @xmath1103 , which is real - valued on @xmath590 , is a constant .
the bishop theorem about antisymmetric subdivisions ( cf . [ g ] , theorem 13.1 ) claims that the maximal sets of antisymmetry @xmath1105 are closed and disjoint , and their union is @xmath1101 .
moreover , if @xmath1106 and @xmath1107 , then @xmath1108 . in particular , if each @xmath1109 is a singleton , then @xmath1110 . for a space @xmath1111 ,
let us denote by @xmath1112 the closure in the @xmath38-norm on compacts in @xmath286 of the algebra @xmath1113 generated by all complex polynomial functions .
note that if any two points in @xmath286 can be separated by a _
real - valued _ polynomial , then bishop s theorem implies that @xmath1114 .
for instance , consider a section @xmath1115 of the complex surface @xmath871 by a totally real subspace @xmath1116 an image of @xmath1117 under a complex transformation @xmath1118 .
since any two points in @xmath1119 , can be separated by a real - valued polynomial , the same property holds for any two points in @xmath1120 , and thus , in @xmath1115 . by the bishop theorem ,
any continuous function @xmath37 on @xmath1115 admits an approximation in the @xmath38-norm on compacts by complex polynomials .
in particular , this conclusion is valid for the real surfaces @xmath1121 and @xmath1122 which have been employed on many occasions . for a compact set @xmath1123 , denote by @xmath1124 the polynomial hull ( closure ) of @xmath1101 .
it consists of all points @xmath1125 in @xmath157 with the property : @xmath1126 for _ any _ complex polynomial @xmath8 .
because for any point @xmath1127 and @xmath1128 , we have @xmath1129 , the polynomial closure @xmath1124 must be contained in @xmath53 .
in fact , the polynomial closure @xmath1124 of a compact @xmath1130 must be a polynomially convex compact set . a theorem by oka and weyl ( cf .
[ g ] , theorem 5.1 ) claims that any complex analytic function , defined in a neighborhood of a compact polynomialy convex set , admits an approximation in the @xmath38-norm on @xmath1101 by complex polynomials .
thus , any analytic function @xmath1131 defined in a neighborhood of @xmath1132 , @xmath1101 being a compact in @xmath871 , can be uniformly approximated on @xmath1124 by complex polynomials .
when dealing with families of functions on non - compact sets , the default topology in the relevant functional spaces is defined by the uniform convergency on compact subsets . in this topology
, the subalgebra @xmath1133 of holomorphic functions is closed in the algebra of all continuous functions @xmath1134 ( cf .
[ gur ] , lemma 11 ) .
in particular , if a sequence of polynomials ( in @xmath5 , and @xmath6 ) is converging in the @xmath38-norm on every compact in @xmath871 , then its limit is a holomorphic function on @xmath871 .
one can employ any expanding family @xmath1135 ( i.e. , @xmath1136 and @xmath1137 ) of polynomially convex compacts @xmath1138 to build a sequence @xmath1139 of polynomials that will approximate a given holomorphic function @xmath1140 .
let @xmath1141 be a subset of @xmath1133 formed by functions on @xmath871 that admit a representation as a series @xmath1142 where each @xmath4 is a linear form in @xmath5 , and @xmath6 .
the series is required to converge uniformly on each compact @xmath1123 ( as we remarked before , any such uniformly converging series produces a holomorphic function on @xmath871 ) .
in fact , one can define a similar set @xmath1143 for any closed @xmath1130 .
due to theorem 11 , any polynomial on @xmath871 is of the form @xmath844 ( which is a special case of @xmath1144 ) .
therefore , when @xmath1145 for a compact @xmath1123 , then @xmath1146 is dense in @xmath1100 as well .
examining @xmath1144 , we observe that if this series converges at a point @xmath1147 it must converge absolutely at any other point @xmath1148 , were the complex number @xmath99 has modulus less than 1 .
for any set @xmath1149 , we denote by @xmath1150 the set @xmath1151 and call it the _ round hull _ of @xmath590 .
consider the set @xmath1152 . because any complex line through the origin that does not belong to the complex cone @xmath318 hits @xmath871 at a pair of antipodal points , @xmath1152 is an open domain in @xmath157 , complementary to the cone ( the origin belongs to @xmath1152 ) , whose boundary contains @xmath871 .
therefore , any function @xmath1153 must be a restriction of a function which is _ analytic _ in @xmath1152 and continuous in @xmath1154 .
i doubt that the converse statement is true .
note that a given function @xmath37 on @xmath871 may have many analytic extensions in @xmath1152 : for example , 1 extends to 1 and to @xmath1155 ( the latter has poles along the complex cone ) .
if a section @xmath1156 by a totally real subspace @xmath1120 is an ellipsoid , then its interior in @xmath1120 coincides with @xmath1157 .
thus , any real function @xmath37 on @xmath1115 that admits a representation as in @xmath1144 must be real analytic in the interior of the ellipsoid .
so , it is represented by its taylor series at the origin ; on the other hand , series @xmath1144 , uniformly converging in the vicinity of the origin , is a form of a very specialized taylor series ( just count the dimensions of the coefficient spaces of each degree to see how special it is ) . since any function from @xmath1133 admits a polynomial approximation on compacts , the subset @xmath1141 is dense in in the space of all holomorphic functions .
so , @xmath1141 is squeezed between the vector space @xmath1133 and its dense subspace @xmath1158 .
it is not even clear whether @xmath1141 is a vector space . to understand the structure of the set @xmath1141 seems to be an interesting and hard problem
unfortunately , our progress towards this goal is minimal .
note that , if a holomorphic function vanishes on a totally real analytic surface @xmath1156 , it must vanish in a neighborhood of @xmath1115 in @xmath871 .
we summarize these observations in the following proposition that , in particular , generalizes the second claim in theorem 3 .
any holomorphic function on complex surface @xmath871 is a limit in the topology of uniform convergence on compacts of polynomial functions in @xmath157 ; the approximating polynomials each admit a representation as in @xmath844 . a subset @xmath1141 of functions that admit a representation as in @xmath1144 is dense in the space of all holomorphic functions @xmath1133 and invariant under multiplications by scalars and the natural @xmath192-action .
each function @xmath37 from @xmath1141 admits a canonic holomorphic extension @xmath1159\ ] ] into the round hull @xmath1152 . here
@xmath1160 and the module of @xmath99 is less than one .
any continuous function @xmath1161 on the totally real surfaces @xmath1162 or @xmath1163 is a limit in the topology of uniform convergence on compacts in @xmath1115 of polynomial functions in @xmath157 . as a result ,
the set @xmath1164 is dense in @xmath1165 .
each function from @xmath1164 admits a similar analytic extension into the open portion of real cone over @xmath1115 that is bounded by @xmath1115 and contains the origin . if @xmath1166 for a function @xmath883 , holomorphic in a neighborhood of @xmath1115 in @xmath871 , then such an @xmath883 is unique .
@xmath107 now consider the vector space @xmath1158 of all polynomial functions restricted to @xmath871 and equipped with the inner product @xmath1167 defined by @xmath879 . if a
_ homogeneous _
polynomial @xmath1168 , then @xmath1169 . an homogeneous polynomial is determined by its restriction to @xmath871 .
evidently , @xmath1170 implies that the restriction of @xmath8 to @xmath864 is zero .
since @xmath1171 is a totally real analytic submanifold , @xmath1172 implies that @xmath8 vanishes in the vicinity of @xmath864 in @xmath871 . by analyticity
, @xmath8 must vanish everywhere in @xmath871 , and hence , @xmath8 is a zero polynomial . as a result ,
@xmath1173 gives rise to a norm on the space of homogeneous polynomials @xmath111 . because any function from @xmath1158 is a restriction of an homogeneous polynomial
, we get a non - degenerated hermitian inner product on the vector space @xmath1158 . in view of theorem 12 and by a similar line of arguments , @xmath1170 implies that @xmath1174 for any complex polynomial @xmath215 .
therefore , being restricted to a space of @xmath66-harmonic polynomials , the inner product @xmath1167 in @xmath879 gives rise to an hermitian structure and an @xmath32-norm @xmath1175 .
in particular , each space @xmath1176 inherits this norm , and @xmath1177 is orthogonal to @xmath1178 , provided @xmath1179 .
consider the vector space @xmath1180 formed by finite sequences of vectors @xmath1181 and its closure @xmath1182 formed by infinite sequences @xmath1181 subject to the condition @xmath1183 .
let @xmath1184 denote the complex hilbert space of @xmath32-integrable functions on the ellipsoid @xmath864 .
every function @xmath1185 defines a unique system of its fourier components " @xmath1186 .
each @xmath1187 is the unique polynomial from @xmath1177 that delivers the minimum @xmath1188 . by theorem 19 ,
any @xmath1189 is a limit in the @xmath38-norm on @xmath864 of harmonic polynomials , it must be also the limit of the same sequence of polynomials viewed as elements of @xmath1184 .
therefore , if @xmath1189 is orthogonal to all the subspaces @xmath1177 , it must be a zero function .
hence , as an element of @xmath1184 , @xmath1190 , and @xmath1191 .
in particular , any @xmath1140 determines a unique system of its harmonics @xmath1192 .
moreover , it is determined by @xmath1193 , and thus , by its harmonics @xmath1194 .
evidently , the sequence of partial sums @xmath1195 } : = \sum_{k = 0}^d f_k \in ker(\delta_q)\}_d$ ] converges in @xmath1184 to @xmath37 . in terms of homogeneous polynomials , we get an analogous sequence @xmath1196 converging in @xmath1184 to @xmath37 .
similar arguments can be applied , instead of the ellipsoid @xmath864 , to any surface @xmath1156 that is a section of @xmath871 by a totally real subspace @xmath1116 . in particular , they are valid for @xmath1197 .
first , we pick a measure @xmath134 on @xmath1115 such that any polynomial @xmath1198 ( when @xmath1115 is compact , one can choose any measure of finite volume ) .
for example , consider the area 2-form @xmath1029 on @xmath1115 induced by the imbedding @xmath1199 and multiply it by the function @xmath1200 $ ] to get the right @xmath134 .
then we define a new inner product by @xmath1201 .
since @xmath1115 is totally real , this will give rise to an hermitian structure in each space @xmath1202 .
notice that the multiplication - by-@xmath66 imbedding @xmath1203 is an isometry .
as before , we form the orthogonal compliments @xmath1204 to the subspaces @xmath1205 ( the only difference is that now the space @xmath119 could be different from the space @xmath1206 of @xmath66-harmonic polynomials ) .
then we use @xmath1207 to construct a hilbert space @xmath1208 .
as before , any continuous function @xmath37 on @xmath1115 will acquire a unique a representation @xmath1190 , where @xmath1209 , and @xmath1210 .
thanks to theorems 7 and 11 , each complex homogeneous polynomial @xmath1211 can be represented by some multipole @xmath1212 . similarly , any real homogeneous polynomial @xmath1213 can be represented by some multipole @xmath1214 .
when @xmath66 is positive definite , these real multipoles are unique .
however , in general , the ambiguity of the multipole representation could cause some trouble .
so , we need to choose the representing multipoles with some care
. due to the embedding @xmath1215 ( see ( 16 ) ) with the image @xmath1216 , the multipole space @xmath1217 acquires a metric @xmath653 induced by the @xmath1218-norm @xmath1219 in @xmath1202 .
the lemma below helps to estimate the size of the fiber @xmath1220 over @xmath1221 in terms of an universal angle @xmath1222 and the @xmath1218-norm of @xmath1223 : consider the distance function @xmath653 on @xmath1217 that is induced by the @xmath1218-norm @xmath1224 in @xmath1202 via the embedding @xmath1225 .
then there exist an angle @xmath1226 so that , for each @xmath1227 , the distance from any @xmath1228 to the zero multipole is at most @xmath1229 , and the diameter of the fiber @xmath1220 is at most @xmath1230 .
* proof * let @xmath1231 denote a unit sphere ( with respect to the @xmath1224-norm ) in @xmath1202 and centered at the origin . because @xmath66 is irreducible , @xmath1232 .
thus , the compact sets @xmath1233 and @xmath1234 are disjoint .
therefore , there is a number @xmath1235 so that the angle between any two vectors @xmath1236 and @xmath1237 is greater than or equal to @xmath1238 .
note that @xmath1233 and @xmath1239 are invariant under the circle action @xmath1240 .
so , @xmath1216 and @xmath1205 also are real cones with their tips at the origin and bases @xmath1241 and @xmath1239 .
we conclude that the angle between any two vectors @xmath1242 and @xmath1243 has the same lower bound @xmath1244 .
now consider an open real cone @xmath1245 comprised of vectors that form an angle @xmath1246 with the subspace @xmath1205 and a complementary cone @xmath1247 .
the argument above shows that @xmath1248 .
hence , the distance from any @xmath1228 to the zero multipole is at most @xmath1229 . as a result ,
the diameter of the fiber @xmath1220 is at most @xmath1230 .
@xmath107 consider a continuous function @xmath1249 and its orthogonal decomposition @xmath1250 , where @xmath1211 .
then , for any choice of the multipoles @xmath1251 , @xmath1252 where @xmath1253 .
@xmath107 it seems to be far from trivial to understand the asymptotic behavior of @xmath1254 as @xmath1255 . perhaps , the lack of understanding of this asymptotics it is the most significant gap in our analysis . to state the last claim in the next theorem ,
we need one technical definition that likely has very little to do with the essence of the statement .
the set @xmath1256 is a complex algebraic variety that is stratified by algebraic sets @xmath1257 which are labeled by various partitions @xmath14 of @xmath1258 .
this labeling is done by attaching the divisor @xmath1259 , or rather the partition @xmath14 of @xmath1258 defined by the multiplicity function of @xmath531 , to each homogeneous polynomial @xmath0 restricted to the cone @xmath318 and viewed as an element of @xmath1204 . in particular ,
when @xmath1260 or @xmath1261 , then @xmath1262 or @xmath1263 , respectively .
the natural partial order among partitions reflects the inclusions of the corresponding strata .
if we delete all the substrata from a given stratum @xmath1264 , we get a pure " stratum that we denote @xmath1265 .
the variety @xmath374 is a whitney stratified space ; as a result , the vicinity of every stratum @xmath1266 has a structure of a bundle whose fiber is a real cone over another stratified space @xmath1267 the link of @xmath1266 .
we will make use of this fact together with another important feature of the stratification @xmath1264 : namely , all the strata of @xmath1267 have even real codimensions .
we say that a parametric curve @xmath1268 \rightarrow v_q^\perp(k ) \approx har_q(k)$ ] is _ tame _ if it consists of a finite number of arcs , each of which has the following property : the interior of each arc is contained in some stratum @xmath1265 .
we say that a continuous function family @xmath1269 is _ tame _ , if for each @xmath275 , the path @xmath1270 is tame .
first , consider the functions @xmath1140 such that , for each @xmath275 , the polynomial @xmath1271 does not belong to the ramification locus @xmath1272 of the map @xmath319 ( the rest of the functions form a complex codimension one subset @xmath1273 ) . over the compliment to the variety @xmath374 ,
the map @xmath319 is a covering map .
thus , for each initial lifting , a deformation @xmath1274 admits a unique lifting to the multipole space , as long as @xmath1275 .
next , for any tame @xmath285-family @xmath1276 , consider the tame curve @xmath1277 and the first arc @xmath1278 in a finite sequence of arcs that form this curve .
there are two possibilities : 1 ) the arc starts at a stratum @xmath1265 and is confined to it for a while , 2 ) the arc starts at a stratum @xmath1265 but moves instantly into an ambient stratum @xmath1279 . in the first case , over @xmath1265 , @xmath319 is a covering map and there is a unique lifting of @xmath1278 extending each lifting @xmath1280 of @xmath1281 . in the second case ,
we claim that , for any lifting @xmath1280 of @xmath1281 , in the vicinity of @xmath1280 the map @xmath319 is onto .
indeed , it is a proper holomorphic map , and thus , its image must be an analytic space ( see [ n ] , theorem 2 , page 129 ) . because @xmath319 is finite , the image of a neighborhood of @xmath1280 under @xmath319 must be of the maximal dimension , and hence , must contain a neighborhood of @xmath1281 . as a result , the set @xmath1282 ( it is a finite graph ) must be present in any neighborhood of @xmath1280 ; so , we can lift @xmath1278 to an arc @xmath1283 that starts at @xmath1280 .
an induction by the number of arcs in the curve @xmath1274 proves the existence of its lifting to the multipole space .
note that an analogous argument fails for @xmath1284 : a finite image of a real analytic set is a real semi - analytic set that can miss the arc @xmath1278 .
therefore , in theorem 22 the lifting property for tame deformations is absent .
the arguments above prove the following theorem : let @xmath12 be an irreducible complex quadratic form , and let @xmath871 be a complex quadratic surface defined by the equation @xmath1285 .
denote by @xmath785 a complex change of coordinates that reduces the form @xmath66 to the sum of squares .
let @xmath1171 be a totally real ellipsoid defined by the equations @xmath1286 and equipped with the measure defined by @xmath879 .
let @xmath37 be an analytic function on @xmath871 .
then there exists sequence of multipoles @xmath1287 such that : * the sequence gives rise , via the maps @xmath1288 , to complex @xmath66-harmonic polynomials @xmath1289 , where the mutually orthogonal polynomials @xmath1290 are uniquely determined by @xmath37 . * as @xmath1291 , the polynomials @xmath1292 converge in the space @xmath1184 to the function @xmath1293 , and therefore , uniquely determine @xmath1140 .
* for a given @xmath37 , there are at most @xmath1294 choices for each multipole @xmath1295 . *
the multipoles @xmath1296 satisfy property @xmath1297 from corollary @xmath1298 .
* for any tame deformation @xmath1299 of the function @xmath1300 , there exists a continuous deformation @xmath1301 of the @xmath1302-representing multipoles , such that @xmath1303 . for functions
@xmath37 and their continuous deformations @xmath1302 outside a subspace @xmath1273 of complex codimension one and for each choice of the appropriate multipoles @xmath1304 , the lifting of the deformation @xmath1305 to the multipole spaces is unique .
@xmath107 similarly , we get let @xmath12 be an irreducible real quadratic form .
let @xmath1306 be a real quadratic surface , equipped a measure @xmath134 for which any polynomial in @xmath5 , and @xmath6 is an @xmath32-integrable function on the surface .
let @xmath37 be an @xmath32-integrable continuous function on @xmath1197 .
then there exists sequence of multipoles @xmath1287 such that : * the sequence gives rise , via the maps @xmath1307 , to real polynomials @xmath1308 , where the mutually orthogonal polynomials @xmath1309 are uniquely determined by @xmath37 ; * as @xmath1291 , the polynomials @xmath1292 converge to @xmath37 in the space @xmath1310 ; * the multipoles @xmath1296 satisfy property @xmath1297 from corollary @xmath1298 .
when @xmath66 is positive - definite , * the mutually orthogonal polynomials @xmath1311 ; * each multipole @xmath1295 is uniquely determined by @xmath37 ; * for any continuous deformation @xmath1312 , of the function @xmath1300 , there exists a unique continuous deformation @xmath1301 of the @xmath1302-representing multipoles , such that @xmath1303 .
@xmath107 _ acknowledgments .
_ i am grateful to jeff weeks for introducing me to the subject .
my conversations with michael shubin about the linear aspects " of this investigation were equally enlightening .
i also would like to thank blaine lawson for explaining to me a few facts and constructions that turned out to be very helpful .
[ b ] bennett , c. et al . , _ first year wilkinson microwave anisotropy probe ( wmap 1 ) observations : preliminary maps and basic results _ , astrophysical journal suppliment series 148 ( 2003 ) , 1 - 27 .
[ chs ] copi c.j . , huterer d. , starkman , g.d . , _ multipole vectors a new representation of the cmb sky and evidence for statistical anisotropy or non - gaussianity at _
@xmath1313 , to appear in phys . rev .
d. ( astro - ph/0310511 ) . | any homogeneous polynomial @xmath0 of degree @xmath1 , being restricted to a unit sphere @xmath2 , admits essentially a unique representation of the form @xmath3 $ ] , where @xmath4 s are linear forms in @xmath5 and @xmath6 and @xmath7 s are real numbers .
the coefficients of these linear forms , viewed as 3d vectors , are called _ multipole _ vectors of @xmath8 . in this paper
we consider similar multipole representations of polynomial and analytic functions on other quadratic surfaces @xmath9 , real and complex . over the complex numbers , the above representation is not unique , although the ambiguity is essentially finite .
we investigate the combinatorics that depicts this ambiguity .
we link these results with some classical theorems of harmonic analysis , theorems that describe decompositions of functions into sums of spherical harmonics .
we extend these classical theorems ( which rely on our understanding of the laplace operator @xmath10 ) to more general differential operators @xmath11 that are constructed with the help of the quadratic form @xmath12 .
then we introduce modular spaces of multipoles .
we study their intricate geometry and topology using methods of algebraic geometry and singularity theory .
the multipole spaces are ramified over vector or projective spaces , and the compliments to the ramification sets give rise to a rich family of @xmath13-spaces , where @xmath14 runs over a variety of modified braid groups . |
extra dimensions at the tev scale @xcite have become a popular feature in nowadays model building .
they can provide possible explanations to some of the present problems within the minimal supersymmetric extension of the standard model ( mssm ) like supersymmetry breaking , the flavour structure , electroweak symmetry breaking , or even unification with gravity .
in particular , one of the most appealing possibilities to break supersymmetry in theories with extra dimensions is the scherk - schwarz ( ss ) mechanism @xcite , based on non - trivial twists that generate a mass term through compactification @xcite .
this mechanism appears to be natural in any locally supersymmetric five - dimensional ( 5d ) theory , since it has been proven that the ss mechanism can be seen as a spontaneous breaking of the @xmath2 via some off - shell auxiliary fields @xcite .
being spontaneous this kind of supersymmetry breaking is quite desirable since it will not generate any quadratic divergences @xcite .
the most common way of compactifying extra - dimensional theories is by using orbifolds @xcite which allow for chirality and partial breaking of supersymmetry ; this is needed because theories in extra dimensions are in general non - chiral and have extended supersymmetry .
orbifolds introduce fixed points in the theory , where translational invariance is broken .
therefore terms localized at those points can arise without spoiling the higher dimensional poincare invariance in the bulk .
this type of localized interactions can be used to break supersymmetry dynamically , e.g. via gaugino condensation , or by any other kind of mechanism that could generate a brane mass term for gravitinos .
the possibility of supersymmetry restoration between the ss mechanism and brane localized supersymmetry breakdown was already proposed in the context of the heterotic horava - witten m - theory @xcite .
also , this kind of interactions have been studied in the past in the context of generalized ss mechanism @xcite , since these terms completely change the spectrum and moreover any fermion mass term localized on a brane will generate discontinuities in the wavefunctions .
the aim of this paper is to generalize previous results @xcite and make a systematic analysis of a very general class of models that could shed light on the mechanism of supersymmetry breaking . in this paper
we will study a 5d theory compactified on @xmath3 with ss supersymmetry breaking plus the most general fermion mass terms on the brane , which do not necessarily have to respect @xmath2 since the latter is broken by the compactification .
we will perform a fully analytic study of both the cp - even and the cp - odd brane mass terms and find out the cases where there are massless states in the spectrum , corresponding to situations where there is cancellation of supersymmetry breaking between the different sources .
we will also show how the discontinuities of the fields are treated consistently and are independent of any regularization of the brane .
we will perform all explicit calculations for the case of gauginos that are symplectic majorana spinors transforming as doublets with respect to @xmath2 .
our results will be straightforwardly generalized to the case of symplectic majorana 5d gravitinos that also transform as doublets of @xmath2 .
all our analytic results have been numerically checked using regularized @xmath1-functions .
the layout of the paper is as follows . in section 2
we will present the model with both the bulk lagrangian plus possible mass terms localized on the brane that will be classified according to its transformation properties under the four - dimensional cp - invariance .
section 3 will be devoted to the full analysis of the cp - even brane whereas section 4 contains the cp - odd case .
finally our conclusions will be drawn on section 5 .
we will consider a five - dimensional model compactified on the orbifold @xmath0 where @xmath4 transforms the fifth coordinate @xmath5 into @xmath6 .
fermions in five dimensions can be described by a pair of symplectic majorana spinors @xmath7 that transform as a doublet under @xmath2 and such that @xmath8 where @xmath9 are the two - component spinors @xmath10 and @xmath11 . under @xmath4
the transformation law of @xmath12 is @xmath13 where @xmath14 is the intrinsic parity of the spinor @xmath12 , the matrix @xmath15 is acting on @xmath2 indices , @xmath16 and we are using the metric @xmath17 . then for a spinor with intrinsic parity @xmath18 it turns out that @xmath19 is an even and @xmath20 an odd two - component spinor
. we will find it convenient to define 4d majorana spinors as @xmath21 although obviously they do not transform irreducibly under @xmath2 nor are they covariant under the 5d lorentz group . under @xmath4 parity
@xmath22 is even and @xmath23 odd .
the lagrangian for @xmath12 in the bulk contains @xmath24 where @xmath25 is the covariant derivative with respect to local lorentz and gauge transformations and summation over the @xmath2 indices is always understood .
the kinetic lagrangian is obtained from the 4d covariant derivative @xmath26 as @xmath27 while the kk - mass bulk term comes from the @xmath28 term in @xmath29 as @xmath30 in the off - shell formulation of 5d supergravity @xcite it was proven that @xmath25 also contains the covariant derivative with respect to local @xmath2 tansformations , @xmath31 , where @xmath32 are auxiliary fields in the minimal supergravity multiplet .
this fact was used in refs .
@xcite to interpret the scherk - schwarz ss breaking of supersymmetry as the background configuration @xmath33 . restoring
the @xmath34-dependence at any stage should be a trivial task . ] .
this gives rise to the ss - lagrangian @xmath35 the bulk lagrangian can then be written as @xmath36 in the brane the different possible @xmath4-even mass terms can be either @xmath37-even or @xmath37-odd .
@xmath37-even mass terms on the brane are provided by the following contributions @xmath38 and @xmath39 or an arbitrary linear combination of ( [ cp1 ] ) and ( [ cp2 ] ) .
we will then consider possible mass terms localized at one of the branes , e.g. @xmath40 , as @xmath41\ ] ] where @xmath42 is an arbitrary real parameter .
@xmath37-odd mass terms on the brane can be provided by the contributions @xmath43 and @xmath44 or any arbitrary linear combination of ( [ nocp1 ] ) and ( [ nocp2 ] ) .
we can then consider the brane lagrangian @xmath45\ ] ] where @xmath42 is an arbitrary real parameter .
five - dimensional theories where brane mass terms ( [ cp1 ] ) and ( [ cp2 ] ) , or ( [ nocp1 ] ) and ( [ nocp2 ] ) , appear can be found in the recent literature . on the other hand solving theories where the brane lagrangians ( [ cpeven ] ) and ( [ cpodd ] ) interfere with the scherk - schwarz breaking has been done only for some cases .
in particular the lagrangian ( [ cpeven ] ) for @xmath46 @xcite and the lagrangian ( [ cpodd ] ) for @xmath47 @xcite . in the following sections we will analyze the general cases both for ( [ cpeven ] ) and ( [ cpodd ] )
the lagrangian in this case is given by ( [ bulk2 ] ) and ( [ cpeven ] ) , i.e. @xmath48\ ] ] the equations of motion ( eom ) from the lagrangian ( [ lageven ] ) are given by @xmath49 we can now decompose the fields @xmath50 in plane waves as @xmath51 where @xmath52 , @xmath53 being the mass eigenvalues and @xmath54 the mass eigenfunctions . in the rest frame @xmath55 eq .
( [ eom0 ] ) is written as @xmath56 where @xmath57 are the mass eigenstates in the rest frame .
we define now majorana spinors as @xmath58 where @xmath59 are constant real spinors and @xmath60 complex functions .
we can then write the eom ( [ eom1 ] ) as @xmath61 where the constant two - component spinors @xmath59 factorize out of the eom ( [ eom2 ] ) and play no role in what follows .
an explicit analytic solution of equations ( [ eom2 ] ) with real functions @xmath62 can be obtained in the region @xmath63 , for initial conditions @xmath64 as @xmath65 \vspace{.25cm}\\ \sin\left[(m_n-\omega)y- \arctan\,t_\rho\left(\lambda\varepsilon(y)\right)\right ] \end{array } \right)\ ] ] where @xmath66 is the sign function and we use the notation , @xmath67 for the typical values of @xmath68 this function is given by : @xmath69 in the region @xmath63 the functions ( [ solutioneven ] ) are periodic provided the condition @xmath70 holds . notice
that supersymmetry can be restored if @xmath71 and @xmath72 are such that @xmath73 , where @xmath74 is any integer number in which case there is always a massless mode in the spectrum .
the solution ( [ solutioneven ] ) near the brane , i.e. for @xmath75 where @xmath76 , is given by @xmath77 from ( [ evenbrane ] ) we see that @xmath78 is an odd function that makes a jump at the brane @xmath79 given by @xmath80 while @xmath81 has a discontinuity on the brane @xmath82 which in general is different from @xmath83 . only for the particular case
@xmath47 is the function @xmath84 continuous .
however one can check that near the brane @xmath85= ( \phi^1_n(0))^2+\rho\,(\phi^2_n(0))^2\ ] ] and so the term in the lagrangian ( [ lageven ] ) coupled to the brane is continuous . in fig .
[ cpevenplot ] we present a numerical solution to eq .
( [ eom2 ] ) obtained by regularizing the @xmath1-function with a gaussian of width @xmath86 ( we have checked our results using other regularizations as well ) .
the solutions precisely coincide with the analytic ones , eqs .
( [ solutioneven ] ) and ( [ periodiceven ] ) .
we display in detail the discontinuities found in eq .
( [ jumpeven ] ) and eq .
( [ disceven ] ) . furthermore we plot in fig .
[ cpevencont ] the combination @xmath87 .
note that the scales used in fig .
[ cpevencont ] and in fig .
[ cpevenplot ] ( right panel ) are identical and the continuity of @xmath87 is explicit . if we replace @xmath88 in ( [ lageven ] ) the solution ( [ solutioneven ] ) can be generalized to one periodic for @xmath89 as @xmath90 \vspace{.25cm}\\ \sin\left[(m_n-\omega)y-\widetilde{\eta}(y ) \right ] \end{array } \right)\ ] ] where @xmath91 is an odd function that makes the prefactor periodic along the whole real axis .
the function @xmath92 is given by @xmath93 notice that the function ( [ eta ] ) has the same shape as @xmath94 where @xmath95 is the staircase function @xmath96 that steps by two units every @xmath97 .
however obviously both functions differ by their derivatives on the brane ( in the special case @xmath98 the two functions are indeed identical , @xmath99 ) .
it is a simple matter to prove that the functions ( [ pereven ] ) are periodic in the real axis @xmath100 provided that condition ( [ periodiceven ] ) holds .
the lagrangian in this case is given by ( [ bulk2 ] ) and ( [ cpodd ] ) , i.e. @xmath101\ ] ] we can then write the eom in the rest frame @xmath55 as @xmath102 where @xmath103 are the mass eigenstates corresponding to the mass eigenvalues @xmath53 .
if we define now the majorana spinors as in ( [ majorana2 ] ) we can write the eom ( [ eomodd1 ] ) as @xmath104 unlike eqs .
( [ eom2 ] ) , eqs .
( [ eomodd2 ] ) do not admit real solutions .
in fact by decomposing @xmath105 eqs .
( [ eomodd2 ] ) lead to @xmath106 eqs .
( [ eomodd ] ) can be solved analytically .
we choose the most general initial conditions consistent with parity ( and up to a global normalization factor ) : @xmath107 in the interval @xmath63 an analytic solution is then given by @xmath108 \vspace{.25cm}\\ \sin\left[(m_n-\omega)y- \arctan\left\ { t_\rho\left(\lambda\varepsilon(y)\right)/\zeta \right\}\right ] \end{array } \right)\ ] ] @xmath109 \vspace{.25 cm } \\
-\sin\left[(m_n+\omega)y -\arctan\left\ { \zeta\,\,t_\rho\left(\lambda\varepsilon(y)\right)\right\}\right ] \end{array } \right)\ ] ] where both @xmath110 and @xmath53 are to be determined by imposing periodicity of the solution
. periodicity condition of ( [ soloddphi ] ) and ( [ soloddchi ] ) is implemented by @xmath111 and the initial condition is fixed by @xmath112 where the last equality is implied by ( [ periododd ] ) and @xmath110 turns out to be independent of @xmath74 .
notice that , unlike the case with @xmath37-even brane masses , when @xmath37-odd brane masses are present supersymmetry restoration is not possible for non - trivial values of the parameters @xmath71 and @xmath72 . the solutions ( [ soloddphi ] ) and ( [ soloddchi ] ) near the brane , @xmath75 where @xmath76 , are given by , @xmath113 @xmath114 from ( [ oddphibrane ] ) and ( [ oddchibrane ] ) we see that @xmath78 and @xmath115 are odd functions that make jumps at the brane @xmath79 , @xmath116 given by @xmath117 @xmath118 while @xmath81 and @xmath119 have discontinuities on the brane @xmath120 @xmath121 only for the particular case @xmath47 are the functions @xmath84 and @xmath122 continuous .
however one can check that near the brane @xmath123= \phi^1_n(0)\,\chi^1_n(0)-\rho\,\phi^2_n(0)\,\chi^2_n(0)\ ] ] and so the term coupled to the brane in the lagrangian ( [ lagodd ] ) is continuous .
if we replace @xmath88 in ( [ lagodd ] ) the solution ( [ soloddphi ] ) and ( [ soloddchi ] ) can be generalized to one periodic for @xmath89 by replacing @xmath124 in the prefactors of ( [ soloddphi ] ) and ( [ soloddchi ] ) and by replacing @xmath125\to & \lim_{\kappa\rightarrow\infty } \sum_{k=-\kappa}^\kappa \arctan\ , \zeta\ , t_\rho[\lambda\varepsilon(y-2\pi k)]\nonumber\\
\arctan\ , \frac{1}{\zeta}\ , t_\rho[\lambda\varepsilon(y)]\to & \lim_{\kappa\rightarrow\infty } \sum_{k=-\kappa}^\kappa \arctan\ , \frac{1}{\zeta}\ , t_\rho[\lambda\varepsilon(y-2\pi k)]\end{aligned}\ ] ] in the corresponding arguments of the trigonometric functions . using ( [ periododd ] ) and ( [ initodd ] ) one can check that the new functions are periodic along the real axis .
as in the case of @xmath37-even mass terms we have checked our solution numerically with a regularized @xmath1-function .
we find perfect agreement with our analytical solution , eqs .
( [ soloddphi])([initodd ] ) .
in this paper we have studied the compactification of a 5d theory on @xmath0 with ss supersymmetry breaking and the most general brane mass term .
we have focussed on the case of bulk symplectic majorana fermions and so our results apply to the case of both gauginos and gravitinos . in all cases fermions obey first order differential equations and brane terms correspond to discontinuities on the fields .
we have studied in full detail both the cp - even and cp - odd brane mass cases .
it has been shown that in the former case there can be a cancellation between the contribution of the standard ss and the brane terms and so supersymmetry can be restored whereas in the latter case this is not possible .
although in most cases there is a discontinuous behaviour for the even fermions near the brane , we have checked that the combination that couples to the brane is continuous throughout the orbifold fixed points .
the generalization of this mechanism to a more general framework with brane fields , or to warped scenarios is left for future investigation .
we would like to thank j. bagger and m. redi for discussions .
one of us ( ad ) wants to thank the aspen center for physics where part of this work was done .
gg thanks the daad for financial support .
99 i. antoniadis , _ phys .
lett . _ * b246 * ( 1990 ) 377 ; i. antoniadis , c. muoz and m. quirs , _ nucl
. phys . _ * b397 * ( 1993 ) 515 ; i. antoniadis and k. benakli , _ phys .
* b326 * ( 1994 ) 69 ; i. antoniadis , k. benakli and m. quirs , _ phys . lett .
_ * b331 * ( 1994 ) 313 ; i. antoniadis and m. quirs , _ phys .
_ * b392 * ( 1997 ) 61 .
j. scherk and j.h .
schwarz , _ phys .
lett . _ * b82 * ( 1979 ) 60 ; _ nucl
. phys . _ * b153 * ( 1979 ) 61 ; p. fayet , _ phys . lett . _
* b159 * ( 1985 ) 121 ; _ nucl . phys . _ * b263 * ( 1986 ) 649 .
a. pomarol and m. quirs , _ phys .
lett . _ * b438 * ( 1998 ) 255 ; i. antoniadis , s. dimopoulos , a. pomarol and m. quirs , _ nucl .
* b544 * ( 1999 ) 503 ; a. delgado , a. pomarol and m. quirs , _ phys .
* d60 * ( 1999 ) 095008 .
r. barbieri , l.j .
hall and y. nomura , _ phys .
_ * d63 * ( 2001 ) 105007 ; n. arkani - hamed , l.j .
hall , y. nomura , d. smith and n. weiner , _ nucl .
_ * b605 * ( 2001 ) 81 ; a. delgado , and m. quirs , _ nucl .
* b607 * ( 2001 ) 99 .
g. von gersdorff and m. quiros , _ phys .
* d65 * ( 2002 ) 064016
. g. von gersdorff , m. quiros and a. riotto , _ nucl . phys . _ * b634 * ( 2002 ) 90 .
z. chacko and m.a .
luty , _ jhep _ * 0105 * ( 2001 ) 067 ; t. kobayashi and k. yoshioka , _ phys .
* 85 * ( 2000 ) 5527 ; d. mart and a. pomarol , _ phys . rev . _
* d64 * ( 2001 ) 105025 ; d.e .
kaplan and n. weiner , ` hep - ph/0108001 ` .
a. delgado , g.v .
gersdorff , p. john and m. quirs , _ phys .
_ * b517 * ( 2001 ) 445 ; r. contino , l. pilo , r. ratazzi and a. strumia , _ jhep _ * 0106 * ( 2001 ) 005 ; r. contino and l. pilo , _ phys . lett . _ * b523 * ( 2001 ) 347 ; y. nomura , ` hep - ph/0105113 ` ; n. weiner , ` hep - ph/0106021 ` ; a. delgado , g.v . gersdorff and m. quirs , _ nucl .
_ * b613 * ( 2001 ) 49 ; v.d .
clemente , s.f . king and d.a .
rayner , _ nucl .
_ * b617 * ( 2001 ) 71 ; r. barbieri , l.j .
hall and y. nomura , _ nucl .
phys . _ * b624 * ( 2002 ) 63 ; v.d .
clemente and y.a .
kubyshin , _ nucl .
phys . _ * b636 * ( 2002 ) 115 ; h. d. kim , phys .
d * 65 * ( 2002 ) 105021
. l. j. dixon , j. a. harvey , c. vafa and e. witten , _ nucl .
phys . _ * b261 * ( 1985 ) 678 ; l. j. dixon , j. a. harvey , c. vafa and e. witten , _ nucl . phys . _
* b274 * ( 1986 ) 285 .
i. antoniadis and m. quiros , _ nucl .
_ * b505 * ( 1997 ) 109 ; i. antoniadis and m. quiros , _ phys .
_ * b416 * ( 1998 ) 327 ; i. antoniadis and m. quiros , nucl .
* 62 * ( 1998 ) 312 .
j. a. bagger , f. feruglio and f. zwirner , _ phys .
* 88 * ( 2002 ) 101601 .
j. bagger , f. feruglio and f. zwirner , _ jhep _ * 0202 * ( 2002 ) 010 ; t. gherghetta and a. riotto , _ nucl .
phys . _ * b623 * ( 2002 ) 97 .
k. a. meissner , h. p. nilles and m. olechowski , ` hep - th/0205166 ` . c. biggio and f. feruglio ,
_ annals phys . _
* 301 * ( 2002 ) 65 ; c. biggio , f. feruglio , a. wulzer and f. zwirner , ` hep - th/0209046 ` . m. zucker , _ nucl . phys . _ * b570 * ( 2000 ) 267 ; _ jhep _ * 08 * ( 2000 ) 016 ; _ phys . rev . _
* d64 * ( 2001 ) 024024 .
t. kugo and k. ohashi , _ prog .
* 104 * ( 2000 ) 835 ; t. kugo and k. ohashi , _ prog .
_ * 105 * ( 2001 ) 323 ; t. fujita , t. kugo and k. ohashi , _ prog . theor .
* 106 * ( 2001 ) 671 ; t. kugo and k. ohashi , _ prog . theor . phys . _ * 108 * , 203 ( 2002 ) . | we have analyzed the interplay between scherk - schwarz supersymmetry breaking and general fermion ( gaugino or gravitino ) mass terms localized on the fixed point branes of the @xmath0 orbifold .
analytic solutions for eigenfunctions and eigenvalues are found in all cases .
all results are checked by numerical calculations that make use of regularized @xmath1-functions .
odd and generically also even fermions are discontinuous at the brane fixed points , but in all cases the combination that couples to the brane is continuous . for cp - even brane mass terms supersymmetry restoration can take place when their effects are cancelled by those of scherk - schwarz compactification .
however such a cancellation can not occur for cp - odd brane mass terms .
october 2002 + |
concurrent stochastic games are played by two players on a finite state space for an infinite number of rounds . in every round
, the two players simultaneously and independently choose moves ( or actions ) , and the current state and the two chosen moves determine a probability distribution over the successor states .
the outcome of the game ( or a _ play _ ) is an infinite sequence of states .
these games were introduced by shapley @xcite , and have been one of the most fundamental and well studied game models in stochastic graph games .
we consider @xmath0-regular objectives specified as parity objectives ; that is , given an @xmath0-regular set @xmath1 of infinite state sequences , player 1 wins if the outcome of the game lies in @xmath1
. otherwise , player 2 wins , i.e. , the game is zero - sum .
the class of concurrent stochastic games subsumes many other important classes of games as sub - classes : ( 1 ) _ turn - based stochastic _ games , where in every round only one player chooses moves ( i.e. , the players make moves in turns ) ; and ( 2 ) _ markov decision processes ( mdps ) _ ( one - player stochastic games ) .
concurrent games and the sub - classes provide a rich framework to model various classes of dynamic reactive systems , and @xmath0-regular objectives provide a robust specification language to express all commonly used properties in verification , and all @xmath0-regular objectives can be expressed as parity objectives .
thus concurrent games with parity objectives provide the mathematical framework to study many important problems in the synthesis and verification of reactive systems @xcite ( see also @xcite ) .
the player-1 _ value _ @xmath2 of the game at a state @xmath3 is the limit probability with which player 1 can ensure that the outcome of the game lies in @xmath1 ; that is , the value @xmath2 is the maximal probability with which player 1 can guarantee @xmath1 against all strategies of player 2 .
symmetrically , the player-2 _ value _
@xmath4 is the limit probability with which player 2 can ensure that the outcome of the game lies outside @xmath1 .
the problem of studying the computational complexity of mdps , turn - based stochastic games , and concurrent games with parity objectives has received a lot of attention in literature .
markov decision processes with @xmath0-regular objectives have been studied in @xcite and the results show existence of pure ( deterministic ) memoryless ( stationary ) optimal strategies for parity objectives and the problem of value computation is achievable in polynomial time .
turn - based stochastic games with the special case of reachability objectives have been studied in @xcite and existence of pure memoryless optimal strategies has been established and the decision problem of whether the value at a state is at least a given rational value lies in np @xmath5 conp . the existence of pure memoryless optimal strategies for turn - based stochastic games with parity objectives was established in @xcite , and again the decision problem lies in np @xmath5 conp .
concurrent parity games have been studied in @xcite and for concurrent parity games optimal strategies need not exist , and @xmath6-optimal strategies ( for @xmath7 ) require both infinite memory and randomization in general , and the decision problem can be solved in pspace .
almost all results in the literature consider the problem of computing values and optimal strategies when the game model is given precisely along with the objective .
however , it is often unrealistic to know the precise probabilities of transition which are only estimated through observation .
since the transition probabilities are not known precisely , an extremely important question is how robust is the analysis of concurrent games and its sub - classes with parity objectives with respect to small changes in the transition probabilities .
this question has been largely ignored in the study of concurrent and turn - based stochastic parity games . in this paper
we study the following problems related to continuity and robustness of values : ( 1 ) _ ( continuity of values ) : _ under what conditions can continuity of the value function be proved for concurrent parity games ; ( 2 ) _ ( robustness of values ) : _ can quantitative bounds be obtained on the difference of the value function in terms of the difference of the transition probabilities ; and ( 3 ) _ ( robustness of optimal strategies ) : _ do optimal strategies of a game remain @xmath6-optimal , for @xmath7 , if the transition probabilities are slightly changed . _ our contributions .
_ our contributions are as follows : we consider _ structurally equivalent _ game structures , where the supports of the transition probabilities are the same , but the precise transition probabilities may differ .
we show the following results for structurally equivalent concurrent parity games : _ quantitative bound .
_ we present a quantitative bound on the difference of the value functions of two structurally equivalent game structures in terms of the difference of the transition probabilities .
we show when the difference in the transition probabilities are small , our bound is asymptotically optimal .
our example to show the matching lower bound is on a markov chain , and thus our result shows that the bound for a markov chain can be generalized to concurrent games
. _ value continuity .
_ we show _ value continuity _ for structurally equivalent concurrent parity games , i.e. , as the difference in the transition probabilities goes to 0 , the difference in value functions also goes to 0 .
we then show that the structural equivalence assumption is necessary : we show a family of markov chains ( that are not structurally equivalent ) where the difference of the transition probabilities goes to 0 , but the difference in the value functions is 1 .
it follows that the structural equivalence assumption is both necessary ( even for markov chains ) and sufficient ( even for concurrent games ) .
it follows from above that our results are both optimal ( quantitative bounds ) as well as tight ( assumption both necessary and sufficient ) .
our result for concurrent parity games is also a significant quantitative generalization of a result for concurrent parity games of @xcite which shows that the set of states with value 1 remains same if the games are structurally equivalent .
we also argue that the structural equivalence assumption is not unrealistic in many cases : a reactive system consists of many state variables , and given a state ( valuation of variables ) it is typically known which variables are possibly updated , and what is unknown is the precise transition probabilities ( which are estimated by observation ) .
thus the system that is obtained for analysis is structurally equivalent to the underlying original system and it only differs in precise transition probabilities . for turn - based stochastic parity games the value continuity and the quantitative bounds are same as for concurrent games .
we also prove a stronger result for structurally equivalent turn - based stochastic games that shows that along with continuity of the value function , there is also robustness property for pure memoryless optimal strategies .
more precisely , for all @xmath7 , we present a bound @xmath8 , such that any pure memoryless optimal strategy in a turn - based stochastic parity game is an @xmath6-optimal strategy in every structurally equivalent turn - based stochastic game such that the transition probabilities differ by at most @xmath9 .
our result has deep significance as it allows the rich literature of work on turn - based stochastic games to carry over robustly for structurally equivalent turn - based stochastic games .
as argued before the model of turn - based stochastic game obtained to analyze may differ slightly in precise transition probabilities , and our results shows that the analysis on the slightly imprecise model using the classical results carry over to the underlying original system with small error bounds .
our results are obtained as follows .
the result of @xcite shows that the value function for concurrent parity games can be characterized as the limit of the value function of concurrent multi - discounted games ( concurrent discounted games with different discount factors associated with every state ) .
there exists bound on difference on value function of discounted games @xcite , however , the bound depends on the discount factor , and in the limit gives trivial bounds ( and in general this approach does not work as value continuity can not be proven in general and the structural equivalence assumption is necessary ) .
we use a classical result on markov chains by friedlin and wentzell @xcite and generalize a result of solan @xcite from markov chains with single discount to markov chains with multi - discounted objective to obtain a bound that is independent of the discount factor for structurally equivalent games .
then the bound also applies when we take the limit of the discount factors , and gives us the desired bound .
our paper is organized as follows : in section [ sec : defn ] we present the basic definitions , in section [ sec : mc ] we consider markov chains with multi - discounted and parity objectives ; in section [ sec : games ] ( subsection [ sec : tb ] ) we prove the results related to turn - based stochastic games ( item ( 2 ) of our contributions ) and finally in subsection [ sec : conc ] we present the quantitative bound and value continuity for concurrent games along with the two examples to illustrate the asymptotic optimality of the bound and the structural equivalence assumption is necessary .
detailed proofs are presented in the appendix .
in this section we define game structures , strategies , objectives , values and present other preliminary definitions
. * probability distributions . * for a finite set @xmath10 , a _ probability distribution _ on @xmath10 is a function @xmath11 $ ] such that @xmath12 .
we denote the set of probability distributions on @xmath10 by @xmath13 . given a distribution @xmath14 , we denote by @xmath15 the _ support _ of the distribution @xmath16 . * concurrent game structures . *
a ( two - player ) _ concurrent stochastic game structure _ @xmath17 consists of the following components . * a finite state space @xmath18 and a finite set @xmath10 of moves ( or actions ) . * two move assignments @xmath19 . for @xmath20 , assignment
@xmath21 associates with each state @xmath22 the nonempty set @xmath23 of moves available to player @xmath24 at state @xmath3 . * a probabilistic transition function @xmath25 , which associates with every state @xmath22 and moves
@xmath26 and @xmath27 a probability distribution @xmath28 for the successor state .
* at every state @xmath29 , player 1 chooses a move @xmath30 , and simultaneously and independently player 2 chooses a move @xmath31 .
the game then proceeds to the successor state @xmath32 with probability @xmath33 , for all @xmath34 . for all states @xmath22 and moves
@xmath35 and @xmath27 , we indicate by @xmath36 the set of possible successors of @xmath3 when moves @xmath37 , @xmath38 are selected .
a _ path _ or a _ play _ of @xmath39 is an infinite sequence @xmath40 of states in @xmath18 such that for all @xmath41 , there are moves @xmath42 and @xmath43 such that @xmath44 .
we denote by @xmath45 the set of all paths .
we denote by @xmath46 the random variable that denotes the @xmath24-th state of a path . for a play @xmath47
, we define @xmath48 to be the set of states that occur infinitely often in @xmath49 . * special classes of games .
* we consider the following special classes of concurrent games . _ turn - based stochastic games .
_ a game structure @xmath39 is _ turn - based stochastic _ if at every state at most one player can choose among multiple moves ; that is , for every state @xmath22 there exists at most one @xmath20 with @xmath50 . _ markov decision processes . _
a game structure is a _
player-1 markov decision process ( mdp ) _
if for all @xmath22 we have @xmath51 , i.e. , only player 1 has choice of actions in the game .
similarly , a game structure is a _
player-2 mdp _ if for all @xmath22 we have @xmath52 .
_ markov chains .
_ a game structure is a markov chain if for all @xmath22 we have @xmath52 and @xmath51 .
hence in a markov chain the players do not matter , and for the rest of the paper a markov chain consists of a tuple @xmath53 where @xmath54 is the probabilistic transition function .
* strategies . *
a _ strategy _ for a player is a recipe that describes how to extend a play .
formally , a strategy for player @xmath55 is a mapping @xmath56 that associates with every nonempty finite sequence @xmath57 of states , representing the past history of the game , a probability distribution @xmath58 used to select the next move .
the strategy @xmath59 can prescribe only moves that are available to player @xmath24 ; that is , for all sequences @xmath60 and states @xmath29 , we require that @xmath61 .
we denote by @xmath62 the set of all strategies for player @xmath55 . given a state @xmath29 and two strategies @xmath63 and @xmath64 , we define @xmath65 to be the set of paths that can be followed by the game , when the game starts from @xmath3 and the players use the strategies @xmath66 and @xmath67 . formally , @xmath68 if @xmath69 and if for all @xmath41 there exist moves @xmath42 and @xmath43 such that ( i ) @xmath70 ; ( ii ) @xmath71 ; and ( iii ) @xmath44 . once the starting state @xmath3 and the strategies @xmath66 and @xmath67 for the two players have been chosen , the probabilities of events are uniquely defined @xcite , where an _ event _ @xmath72 is a measurable set of paths . for an event @xmath72 ,
we denote by @xmath73 the probability that a path belongs to @xmath74 when the game starts from @xmath3 and the players use the strategies @xmath66 and @xmath67 . *
classification of strategies .
* we consider the following special classes of strategies . _
( pure ) . _
a strategy @xmath75 is _ pure ( deterministic ) _ if for all @xmath57 there exists @xmath76 such that @xmath77 .
thus , deterministic strategies are equivalent to functions @xmath78 . _
( finite - memory ) .
_ strategies in general are _ history - dependent _ and can be represented as follows : let @xmath79 be a set called _ memory _ to remember the history of plays ( the set @xmath79 can be infinite in general ) .
a strategy with memory can be described as a pair of functions : ( a ) a _ memory update _ function @xmath80 , that given the memory @xmath79 with the information about the history and the current state updates the memory ; and ( b ) a _ next move
_ function @xmath81 that given the memory and the current state specifies the next move of the player . a strategy is _ finite - memory _ if the memory @xmath79 is finite . _
( memoryless ) .
_ a _ memoryless _ strategy is independent of the history of play and only depends on the current state .
formally , for a memoryless strategy @xmath75 we have @xmath82 for all @xmath22 and all @xmath83 . thus memoryless strategies are equivalent to functions @xmath84 . _
( pure memoryless ) . _
a strategy is _ pure memoryless _ if it is both pure and memoryless .
pure memoryless strategies neither use memory , nor use randomization and are equivalent to functions @xmath85 .
* qualitative objectives . *
we specify _ qualitative _ objectives for the players by providing the set of _ winning plays _
@xmath86 for each player . in this paper
we study only zero - sum games @xcite , where the objectives of the two players are complementary .
a general class of objectives are the borel objectives @xcite .
borel objective _ @xmath87 is a borel set in the cantor topology on @xmath88 . in this paper
we consider _
@xmath0-regular objectives _ , which lie in the first @xmath89 levels of the borel hierarchy ( i.e. , in the intersection of @xmath90 and @xmath91 ) @xcite .
all @xmath0-regular objectives can be specified as parity objectives , and hence in this work we focus on parity objectives , and they are defined as follows . _ parity objectives . _ for @xmath92 , we let @xmath93 = { \{c , c+1 , \ldots , d\}}$ ] .
let @xmath94 $ ] be a function that assigns a _ priority _ @xmath95 to every state @xmath22 , where @xmath96 .
the _ even parity objective _ requires that the minimum priority visited infinitely often is even .
formally , the set of winning plays is defined as @xmath97 . * quantitative objectives . *
_ quantitative _ objectives are measurable functions @xmath98
. we will consider _ multi - discounted _ objective functions , as there is a close connection established between concurrent games with multi - discounted objectives and concurrent games with parity objectives . given a concurrent game structure with state space @xmath18 ,
let @xmath99 be a _ discount vector _ that assigns for all @xmath22 a discount factor @xmath100 ( unless otherwise mentioned we will always consider discount vectors @xmath99 such that for all @xmath22 we have @xmath100 ) .
let @xmath101 be a reward function that assigns a real - valued reward @xmath102 to every state @xmath22 .
the multi - discounted objective function @xmath103 maps every path to the mean - discounted reward of the path .
formally , the function is defined as follows : for a path @xmath104 we have @xmath105 also note that a parity objective @xmath1 can be intepreted as a function @xmath106 by simply considering the characteristic function that assigns 1 to paths that belong to @xmath1 and 0 otherwise .
* values , optimality , @xmath6-optimality . * given an objective @xmath1 which is a measurable function @xmath107 , we define the _ value _ for player 1 of game @xmath108 with objective @xmath1 from the state @xmath22 as @xmath109 i.e. , the value is the maximal expectation with which player 1 can guarantee the satisfaction of @xmath1 against all player 2 strategies .
given a player-1 strategy @xmath66 , we use the notation @xmath110 a strategy @xmath66 for player 1 is _ optimal _ for an objective @xmath1 if for all states @xmath22 , we have @xmath111 for @xmath112 , a strategy @xmath66 for player 1 is _ @xmath6-optimal _ if for all states @xmath22 , we have @xmath113 the notion of values , optimal and @xmath6-optimal strategies for player 2 are defined analogously .
the following theorem summarizes the results in literature related to determinacy and memory complexity of concurrent games and its sub - classes for parity and multi - discounted objectives .
[ thrm_lit1 ] the following assertions hold : 1 . _
( determinacy @xcite)_. for all concurrent game structures and for all parity and multi - discounted objectives @xmath1 we have @xmath114 2 . _
( memory complexity ) .
_ for all concurrent game structures and for all multi - discounted objectives @xmath1 , randomized memoryless optimal strategies exist @xcite . for all turn - based stochastic game structures and for all multi - discounted objectives @xmath1
, pure memoryless optimal strategies exist @xcite . for all turn - based stochastic game strucutures and for all parity objectives @xmath1 , pure memoryless optimal strategies exist @xcite . in general optimal strategies
need not exist in concurrent games with parity objectives , and @xmath6-optimal strategies , for @xmath7 , need both randomization and infinite memory in general @xcite .
the results of @xcite established that the value of concurrent games with certain special multi - discounted objectives can be characterized as valuations of quantitaive discounted @xmath115-calculus formula . in the limit
, the value function of the discounted @xmath115-calculus formula characterizes the value function of concurrent games with parity objectives .
an elegant interpretation of the result was given in @xcite , and from the interpretation we obtain the following theorem .
[ thrm_lit2 ] let @xmath108 be a concurrent game structure with a parity objective @xmath1 defined by a priority function @xmath116 .
let @xmath117 be a reward function that assigns reward 1 to even priority states and reward 0 to odd priority states .
then there exists an order @xmath118 on the states ( where @xmath119 ) dependent only on the priority function @xmath116 such that @xmath120 in other words , if we consider the value function @xmath121 with the multi - discounted objective and take the limit of the discount factors to 1 in the order of the states we obtain the value function for the parity objective
. we now present notions related to _ structure equivalent _ game structures and distances .
* structure equivalent game structures .
* given two game structures @xmath122 and @xmath123 on the same state and action space , with different transition function , we say that @xmath124 and @xmath125 are _ structure equivalent _
( denoted @xmath126 ) if for all @xmath22 and all @xmath127 and @xmath27 we have @xmath128 .
similarly , two markov chains @xmath129 and @xmath130 are structurally equivalent ( denoted @xmath131 ) if for all @xmath29 we have @xmath132 . for a game structure @xmath108 ( resp .
markov chain @xmath108 ) , we denote by @xmath133\!]_\equiv}}$ ] the set of all game structures ( resp .
markov chains ) that are structurally equivalent to @xmath108 .
* ratio and absolute distances . * given two game structures @xmath134 and @xmath123 , the _ absolute distance _ of the game structures is maximum absolute difference in the transition probabilities .
formally , @xmath135 .
the absolute distance for two markov chains @xmath129 and @xmath130 is @xmath136 .
we now define the ratio distance between two structurally equivalent game structures and markov chains .
let @xmath137 and @xmath138 be two structurally equivalent game structures .
the _ ratio _
distance is defined on the ratio of the transition probabilities .
formally , @xmath139 the ratio distance between two structurally equivalent markov chains @xmath137 and @xmath138 is @xmath140 .
* remarks about the distance functions .
* we first remark that the ratio distance is not necessarily a metric .
consider the markov chain with state space @xmath141 and let @xmath142 .
for @xmath143 consider the transition functions @xmath144 such that @xmath145 , for all @xmath146 .
let @xmath147 be the markov chain with transition function @xmath144 .
then we have @xmath148 and @xmath149 , and hence @xmath150 .
the above example is from @xcite .
also note that @xmath151 is only defined for structurally equivalent game structures , and without the assumption @xmath151 is @xmath152 .
we also remark that the absolute distance that measures the difference in the transition probabilities is the most intuitive measure for the difference of two game structures .
[ prop_dist ] let @xmath137 be a game structure ( resp .
markov chain ) such that the minimum positive transition probability is @xmath153 . for all game structures ( resp .
markov chains ) @xmath154\!]_\equiv}}$ ] we have @xmath155 .
* notation for fixing strategies . * given a concurrent game structure @xmath156 , let @xmath66 be a randomized memoryless strategy .
fixing the strategy @xmath66 in @xmath108 we obtain a player-2 mdp , denoted as @xmath157 , defined as follows : ( 1 ) the state space is @xmath18 ; ( 2 ) for all @xmath22 we have @xmath158 ( hence it is a player-2 mdp ) ; ( 3 ) the new transition function @xmath159 is defined as follows : for all @xmath22 and all @xmath160 we have @xmath161 .
similarly if we fix a randomized memoryless strategy @xmath66 in an mdp @xmath108 we obtain a markov chain , denoted as @xmath157 .
the following proposition is straightforward to verify from the definitions .
[ prop_stra_fix ] let @xmath137 and @xmath138 be two concurrent game structures ( resp .
mdps ) that are structurally equivalent .
let @xmath66 be a randomized memoryless strategy
. then @xmath162 and @xmath163 .
in this section we consider markov chains with multi - discounted and parity objectives .
we present a bound on the difference of value functions of two structurally equivalent markov chains that is dependent on the distance between the markov chains and is _ independent _ of the discount factors .
the result for parity objectives is then a consequence of our result for multi - discounted objectives and theorem [ thrm_lit2 ] .
our result crucially depends on a result of friedlin and wentzell for markov chains and we present this result below , and then use it to obtain the main result of the section . *
result of friedlin and wentzell .
* let @xmath53 be a markov chain and let @xmath164 be the initial state .
let @xmath165 be a proper subset of @xmath18 and let us denote by @xmath166 the first hitting time to the set @xmath167 of states ( or the first exit time from set @xmath168 ) ( recall that @xmath169 is the random variable to denote the @xmath170-th state of a path ) .
let @xmath171 denote the set of all functions from @xmath168 to @xmath18 .
for every @xmath172 we define a directed graph @xmath173 where @xmath174 iff @xmath175 .
let @xmath176 if the directed graph @xmath177 has no directed cycles ( i.e. , @xmath177 is a directed acyclic graph ) ; and @xmath178 otherwise . observe that since @xmath179 is a function , for every @xmath180 there is exactly one path that starts at @xmath3 .
for every @xmath181 and every @xmath34 , let @xmath182 if the directed path that leaves @xmath3 in @xmath177 reaches @xmath32 , otherwise @xmath183 .
we now state a result that can be obtained as a special case of the result from friedlin and wentzell @xcite .
below we use the formulation of the result as presented in @xcite ( lemma 2 of @xcite ) .
[ thrm - fw ] let @xmath53 be a markov chain , and let @xmath184 be a proper subset of @xmath18 such that @xmath185 for every @xmath181 ( i.e. , from all @xmath181 with positive probability the first hitting time to the complement set is finite ) .
then for every initial state @xmath186 and for every @xmath187 we have @xmath188 in other words , the probability that the exit state is @xmath32 when the starting state is @xmath189 is given by the expression on the right hand side ( very informally the right hand side is the normalized polynomial expression for exit probabilities ) .
* value function difference for markov chains .
* we will use the result of theorem [ thrm - fw ] to obtain bounds on the value functions of markov chains .
we start with the notion of mean - discounted time . * mean - discounted time . * given a markov chain @xmath53 and a discount vector @xmath99 , we define for every state @xmath22 , the _ mean - discounted time _
the process is in the state @xmath3 .
we first define the mean - discounted time function @xmath190 that maps every path to the mean - discounted time that the state @xmath3 is visited , and the function is formally defined as follows : for a path @xmath104 we have @xmath191 where @xmath192 is the indicator function .
the expected mean - discounted time function for a markov chain @xmath108 with transition function @xmath16 is defined as follows : @xmath193 $ ] , i.e. , it is the expected mean - discounted time for @xmath3 when the starting state is @xmath189 , where the expectation measure is defined by the markov chain with transition function @xmath16 .
we now present a lemma that shows the value function for multi - discounted markov chains can be expressed as ratio of two polynomials ( the result is obtained as a simple extension of a result of solan @xcite ) . [ lemm : ratio ] for markov chains defined on state space @xmath18 , for all initial states @xmath164 , for all states @xmath3 , for all discount vectors @xmath99 , there exists two polynomials @xmath194 and @xmath195 in @xmath196 variables @xmath197 , where @xmath198 such that the following conditions hold : 1 .
the polynomials have degree at most @xmath199 with non - negative coefficients ; and 2 .
for all transition functions @xmath16 over @xmath18 we have @xmath200 , where @xmath201 , @xmath202 and @xmath203 denote the values of the function @xmath204 and @xmath205 such that all the variables @xmath197 is instantiated with values @xmath206 as given by the transition function @xmath16 . _
( sketch ) .
_ we present a sketch of the proof ( details in appendix ) .
fix a discount vector @xmath99 .
we construct a markov chain @xmath207 as follows :
@xmath208 , where @xmath209 is a copy of states of @xmath18 ( and for a state @xmath22 we denote its corresponding copy as @xmath189 ) ; and the transition function @xmath210 is defined below 1 .
@xmath211 for all @xmath212 ( i.e. , all copy states are absorbing ) ; 2 . for @xmath22
we have @xmath213 i.e. , it goes to the copy with probability @xmath214 , it follows the transition @xmath16 in the original copy with probabilities multiplied by @xmath215 .
we first show that for all @xmath164 and @xmath3 we have @xmath216 ; i.e. , the expected mean - discounted time in @xmath3 when the original markov chain starts in @xmath164 is the probability in the markov chain @xmath217 that the first hitting state out of @xmath18 is the copy @xmath189 of the state @xmath3 .
the claim is easy to verify as both @xmath218 and @xmath219 are the unique solution of the following system of linear equations : for all @xmath34 we have @xmath220 we now claim that @xmath221 for all @xmath222 .
this follows since for all @xmath22 we have @xmath223 and since @xmath224 we have @xmath225 .
now we observe that we can apply theorem [ thrm - fw ] on the markov chain @xmath207 with @xmath18 as the set @xmath168 of states of theorem [ thrm - fw ] , and obtain the result . indeed the terms @xmath226 and @xmath227 are independent of @xmath16 , and the two products of equation ( [ eq1 ] ) each contains at most @xmath199 terms of the form @xmath228 for @xmath229 .
thus the desired result follows .
[ lemm : poly ] let @xmath230 be a polynomial function with non - negative coefficients of degree at most @xmath170 .
let @xmath7 and @xmath231 be two non - negative vectors such that for all @xmath232 we have @xmath233 .
then we have @xmath234 . [ lemm_mc_multi ]
let @xmath235 and @xmath236 be two structurally equivalent markov chains .
for all non - negative reward functions @xmath101 such that the reward function is bounded by 1 , for all discount vectors @xmath99 , for all @xmath22 we have @xmath237 ; i.e. , the absolute difference of the value functions for the multi - discounted objective is bounded by @xmath238 .
the proof of lemma [ lemm_mc_multi ] uses lemma [ lemm : ratio ] and lemma [ lemm : poly ] and is presented in the appendix .
[ thrm_mc_multi ] let @xmath235 and @xmath236 be two structurally equivalent markov chains .
let @xmath239 be the minimum positive transition probability in @xmath137 .
the following assertions hold : 1 . for
all non - negative reward functions @xmath101 such that the reward function is bounded by 1 , for all discount vectors @xmath99 , for all @xmath22 we have @xmath240 & \leq & \displaystyle ( 1+{\varepsilon}_a)^{2\cdot |s|}-1 \end{array}\ ] ] 2 . for all parity objectives @xmath1 and for all @xmath22 we have @xmath241 where @xmath242 and @xmath243 .
the first part follows from lemma [ lemm_mc_multi ] and proposition [ prop_dist ] .
the second part follows from part 1 , the fact the value function for parity objectives is obtained as the limit of multi - discounted objectives ( theorem [ thrm_lit2 ] ) , and the fact the bound for part 1 is independent of the discount factors ( hence independent of taking the limit ) .
* remark on structural assumption in the proof .
* the result of the previous theorem depends on the structural equivalence assumption in two crucial ways .
they are as follows : ( 1 ) proposition [ prop_dist ] that establishes the relation of @xmath151 and @xmath244 only holds with the assumption of structural equivalence ; and ( 2 ) without the structural equivalence assumption @xmath151 is @xmath152 , and hence without the assumption the bound of the previous theorem is @xmath152 , which is a trivial bound .
we will later show ( in example [ examp1 ] ) that the structural equivalence assumption is necessary .
in this section we show two results : first we show robustness of strategies and present quantitative bounds on value functions for turn - based stochastic games and then we show continuity for concurrent parity games . in this section
we present quantitative bounds for robustness of optimal strategies in structurally equivalent turn - based stochastic games . for every @xmath7 , we present a bound @xmath8 , such that if the distance of the structurally equivalent turn - based stochastic games differs by at most @xmath9 , then any pure memoryless optimal strategy in one game is @xmath6-optimal in the other .
the result is first shown for mdps and then extended to turn - based stochastic games ( both proofs are in the appendix ) . [ thrm_tb_stochastic ]
let @xmath137 be a turn - based stochastic game such that the minimum positive transition probability is @xmath153 .
the following assertions hold : 1 . for all turn - based stochastic games @xmath154\!]_\equiv}}$ ] , for all parity objectives @xmath1 and for all @xmath22 we have @xmath245 & \leq & \displaystyle \bigg(1+\frac{{\mathit{dist}_a}(g_1,g_2)}{\eta}\bigg)^{2\cdot |s|}-1 \end{array}\ ] ] 2 . for @xmath7 ,
let @xmath246 . for all @xmath154\!]_\equiv}}$
] such that @xmath247 , for all parity objectives @xmath1 , every pure memoryless optimal strategy @xmath66 in @xmath137 is an @xmath6-optimal strategy in @xmath138 . in this section
we show value continuity for structurally equivalent concurrent parity games , and show with an example on markov chains that the continuity property breaks without the structural equivalence assumption . finally with an example on markov chains we show the our quantitative bounds are asymptotically optimal for small distance values .
we start with a lemma for mdps .
[ lemm_mdp_multi ] let @xmath137 and @xmath138 be two structurally equivalent mdps .
let @xmath239 be the minimum positive transition probability in @xmath137 .
for all non - negative reward functions @xmath101 such that the reward function is bounded by 1 , for all discount vectors @xmath99 , for all @xmath22 we have @xmath248 & \leq & \displaystyle \bigg(1+\frac{{\mathit{dist}_a}(g_1,g_2)}{\eta}\bigg)^{2\cdot |s|}-1 \end{array}\ ] ] the main idea of the proof of the above lemma is to fix a pure memoryless optimal strategy and then use the results for markov chains . using the same proof idea , along with randomized memoryless optimal strategies for concurrent game structures and the above lemma
, we obtain the following lemma ( the result is identical to the previous lemma , but for concurrent game structures instead of mdps ) .
[ lemm_conc_multi ] let @xmath137 and @xmath138 be two structurally equivalent concurrent game structures .
let @xmath239 be the minimum positive transition probability in @xmath137 .
for all non - negative reward functions @xmath101 such that the reward function is bounded by 1 , for all discount vectors @xmath99 , for all @xmath22 we have @xmath248 & \leq & \displaystyle \bigg(1+\frac{{\mathit{dist}_a}(g_1,g_2)}{\eta}\bigg)^{2\cdot |s|}-1 \end{array}\ ] ] we now present the main theorem that depends on lemma [ lemm_conc_multi ] . [ thrm_conc_multi ] let @xmath137 and @xmath138 be two structurally equivalent concurrent game structures .
let @xmath239 be the minimum positive transition probability in @xmath137 .
for all parity objectives @xmath1 and for all @xmath22 we have @xmath248 & \leq & \displaystyle \bigg(1+\frac{{\mathit{dist}_a}(g_1,g_2)}{\eta}\bigg)^{2\cdot |s|}-1 \end{array}\ ] ] the result follows from theorem [ thrm_lit2 ] , lemma [ lemm_conc_multi ] and the fact that the bound of lemma [ lemm_conc_multi ] are independent of the discount factors and hence independent of taking the limits . in the following theorem
we show that for structurally equivalent game structures , for all parity objectives , the value function is continuous in the absolute distance between the game structures .
we have already remarked ( after theorem [ thrm_mc_multi ] ) that the structural equivalence assumption is required in our proofs , and we show in example [ examp1 ] that this assumption is necessary .
[ thrm_val_con ] for all concurrent game structures @xmath137 , for all parity objectives @xmath1 @xmath249\!]_\equiv } } , { \mathit{dist}_a}(g_1,g_2 ) \leq { \varepsilon } } \ \
\sup_{s\in s }
let @xmath153 be the minimum positive transition probability in @xmath137 .
by theorem [ thrm_conc_multi ] we have @xmath250\!]_\equiv } } , { \mathit{dist}_a}(g_1,g_2 ) \leq { \varepsilon } } \
\sup_{s\in s } \lim_{{\varepsilon}\to 0 } \bigg(1+\frac{{\varepsilon}}{\eta}\bigg)^{2\cdot |s|}-1\ ] ] the above limit equals to 0 , and the desired result follows . [ examp1 ] in this example we show that in theorem [ thrm_val_con ] the structural equivalence assumption is necessary , and thereby show that the result is tight .
we show an markov chain @xmath137 and a family of markov chains @xmath251 , for @xmath7 , such that @xmath252 ( but @xmath137 is not structurally equivalent to @xmath251 ) with a parity objective @xmath1 and we have @xmath253 .
the markov chains @xmath137 and @xmath251 are defined over the state space @xmath254 , and in @xmath137 both states have self - loops with probability 1 , and in @xmath251 the self - loop at @xmath164 has probability @xmath255 and the transition probability from @xmath164 to @xmath189 is @xmath6 ( see fig [ figure : buchi - lim ] in appendix ) . clearly , @xmath256 .
the parity objective @xmath1 requires to visit the state @xmath189 infinitely often ( i.e. , assign priority 2 to @xmath189 and priority 1 to @xmath164 ) .
then we have @xmath257 as the state @xmath164 is never left , whereas in @xmath251 the state @xmath189 is the only closed recurrent set of the markov chain and hence reached with probability 1 from @xmath164 .
hence @xmath258 .
it follows that @xmath259 .
we now show that our quantitative bound for the value function difference is asymptotically optimal for small distances .
let us denote the absolute distance as @xmath6 , and the quantitative bound we obtain in theorem [ thrm_conc_multi ] is @xmath260 , and if @xmath6 is small , then we obtain the following approximate bound @xmath261 we now illustrate with an example ( on structurally equivalent markov chains ) where the difference in the value function is @xmath262 , for small @xmath6 . consider the markov chain defined on state space @xmath263 as follows : states @xmath164 and @xmath264 are absorbing ( states with self - loops of probability 1 ) and for a state @xmath265 we have @xmath266 ; and @xmath267 ; i.e. , we have a markov chain defined on a line from @xmath268 to @xmath269 ( with @xmath268 and @xmath269 absorbing states ) and the chain moves towards @xmath268 with probability @xmath270 and towards @xmath269 with probability @xmath271 ( see fig [ figure : asym ] with complete details in appendix ) .
our goal is to estimate the probability to reach the state @xmath164 , and let @xmath272 denote the probability to reach @xmath164 from the starting state @xmath273 .
we show ( details in appendix ) that if @xmath274 , then @xmath275 and for @xmath276 , such that @xmath6 is close to 0 , we have @xmath277 .
observe that the markov chain obtained for @xmath274 and @xmath278 are structurally equivalent .
thus the desired result follows .
in this work we studied the robustness and continuity property of concurrent and turn - based stochastic parity games with respect to small imprecision in the transition probabilities .
we presented ( i ) quantitative bounds on difference of the value functions and proved value continuity for concurrent parity games under the structural equivalence assumption , and ( ii ) showed robustness of all pure memoryless optimal strategies for structurally equivalent turn - based stochastic parity games .
we also showed that the structural equivalence assumption is necessary and that our quantitative bounds are asymptotically optimal for small imprecision .
we believe our results will find applications in robustness analysis of various other classes of stochastic games . 10 m. abadi , l. lamport , and p. wolper .
realizable and unrealizable specifications of reactive systems . in _ icalp89 _ , lncs 372 , pages 117 .
springer , 1989 .
r. alur , t.a .
henzinger , and o. kupferman .
alternating - time temporal logic .
, 49:672713 , 2002 .
k. chatterjee , l. de alfaro , and t.a .
the complexity of quantitative concurrent parity games . in _ soda06 _ , pages 678687 .
acm - siam , 2004 .
k. chatterjee , t.a .
henzinger , and m. jurdziski .
games with secure equilibria . in _ lics04 _ , pages 160169 .
ieee , 2004 .
k. chatterjee , m. jurdziski , and t.a .
henzinger . quantitative stochastic parity games . in _ soda04 _ , pages 121130 .
siam , 2004 .
a. church .
logic , arithmetic , and automata . in _ proceedings of the international congress of mathematicians _ , pages 2335 .
institut mittag - leffler , 1962 .
a. condon .
the complexity of stochastic games .
, 96(2):203224 , 1992 . c. courcoubetis and m. yannakakis .
the complexity of probabilistic verification .
, 42(4):857907 , 1995 .
l. de alfaro . .
phd thesis , stanford university , 1997 .
l. de alfaro and t.a .
concurrent omega - regular games . in _
lics00 _ , pages 141154 .
ieee , 2000 .
l. de alfaro , t.a .
henzinger , and r. majumdar .
discounting the future in systems theory . in _ icalp03 _ , lncs 2719 , pages 10221037 .
springer , 2003 .
l. de alfaro and r. majumdar . quantitative solution of omega - regular games . in _ stoc01 _ , pages 675683 .
acm press , 2001 . c. derman . .
academic press , 1970 .
d.l . dill . .
the mit press , 1989 .
k. etessami and m. yannakakis .
recursive concurrent stochastic games . in _
icalp06 ( 2 ) _ , lncs 4052 , springer , pages 324335 , 2006 .
j. filar and k. vrieze . .
springer - verlag , 1997 .
m. i. friedlin and a. d. wentzell . .
springer , 1984 .
h. gimbert and w. zielonka .
discounting infinite games but how and why ?
, 119(1):39 , 2005 . a. kechris .
springer , 1995 .
the determinacy of blackwell games .
, 63(4):15651581 , 1998 .
a. pnueli and r. rosner . on the synthesis of a reactive module . in _
popl89 _ , pages 179190 .
acm press , 1989 .
raghavan and j.a .
algorithms for stochastic games a survey .
, 35:437472 , 1991 .
p.j . ramadge and w.m .
supervisory control of a class of discrete - event processes .
, 25(1):206230 , 1987 .
stochastic games . , 39:10951100 , 1953 .
e. solan .
continuity of the value of competitive markov decision processes .
, 16:831845 , 2003 .
w. thomas .
automata on infinite objects . in j.
van leeuwen , editor , _ handbook of theoretical computer science _ ,
volume b , pages 133191 .
elsevier , 1990 .
. automatic verification of probabilistic concurrent finite - state systems . in _ focs85 _ , pages 327338 .
ieee computer society press , 1985 .
perfect - information stochastic parity games . in _ fossacs04 _ , pages 499513 .
lncs 2987 , springer , 2004 .
_ ( of proposition [ prop_dist ] ) . _ consider @xmath22 , @xmath279 , and @xmath280
. then we have the following two inequalities : we consider @xmath281 , and the argument for @xmath282 is symmetric .
we consider @xmath281 and if @xmath283 , then @xmath284 , and otherwise we have the following inequality : @xmath285 it follows that in both cases we have @xmath286 .
the desired result follows from the above inequalities .
we now present the proof of lemma [ lemm : ratio ] which is obtained as a simple extension of a result of solan @xcite . _
( of lemma [ lemm : ratio ] ) .
_ fix a discount vector @xmath99 .
we construct a markov chain @xmath207 as follows : @xmath208 , where @xmath209 is a copy of states of @xmath18 ( and for a state @xmath22 we denote its corresponding copy as @xmath189 ) ; and the transition function @xmath210 is defined below 1 .
@xmath211 for all @xmath212 ( i.e. , all copy states are absorbing ) ; 2 . for @xmath22
we have @xmath213 i.e. , it goes to the copy with probability @xmath214 , it follows the transition @xmath16 in the original copy with probabilities multiplied by @xmath215 .
we first show that for all @xmath164 and @xmath3 we have @xmath287 i.e. , the expected mean - discounted time in @xmath3 when the original markov chain starts in @xmath164 is the probability in the markov chain @xmath217 that the first hitting state out of @xmath18 is the copy @xmath189 of the state @xmath3 .
the claim is easy to verify as both @xmath218 and @xmath219 are the solutions of the following system of linear equations @xmath288 the fact that @xmath218 is the solution of the above equation follows from the results of discounted reward markov chains ( detailed proofs with uniform discount factor for mdps is available in @xcite ( e.g. , equation 2.15 of @xcite ) , and specialization to markov chains and generalization to discount factor attached to every state is straightforward ) .
the fact that @xmath219 is the solution of the above equation follows from the results of characterization of hitting time for transient markov chains ( see @xcite for details ) .
also the above system of linear equations has a unique solution .
the uniqueness of the solution follows from the fact that this is a contraction mapping , and the proof is as follows : let @xmath289 and @xmath290 be two solutions of the system .
we chose @xmath291 such that @xmath292 , i.e. , @xmath293 is a state that maximizes the difference of the two solutions .
let @xmath294 .
as @xmath295 and @xmath296 are solutions of the above system we have by the triangle inequality @xmath297 & \leq & \displaystyle \eta \cdot \sum_{t\in s } { \lambda}(t ) \cdot { \delta}(s_0)(t ) \leq \eta \cdot \max_{t\in s } { \lambda}(t ) \cdot \sum_{t \in s}{\delta}(s_0)(t ) . \end{array}\ ] ] since @xmath298 , it follows that @xmath299 .
since @xmath300 it follows that we must have @xmath301 and hence the two solutions must coincide .
we now claim that @xmath221 for all @xmath222 .
this follows since for all @xmath22 we have @xmath223 and since @xmath224 we have @xmath225 .
now we observe that we can apply theorem [ thrm - fw ] on the markov chain @xmath207 with @xmath18 as the set @xmath168 of states of theorem [ thrm - fw ] , and obtain the result . indeed the terms @xmath226 and @xmath227 are independent of @xmath16 , and the two products of equation ( [ eq1 ] ) each contains at most @xmath199 terms of the form @xmath228 for @xmath229 .
thus the desired result follows .
we now illustrate the construction of lemma [ lemm : ratio ] with the aid of some examples .
consider the markov chain @xmath108 with states @xmath3 and @xmath32 such that @xmath32 is absorbing and the transition from @xmath3 to @xmath32 has probability 1 , and let the discount factor be @xmath302 for all states .
the markov chain @xmath108 along with @xmath303 is shown in fig .
[ fig : illus_1 ] .
if we start at @xmath3 , the mean - discounted time at @xmath32 is given by @xmath304 in the markov chain @xmath303 , the probability to reach @xmath32 from @xmath3 is @xmath302 , and once @xmath32 is reached the exit state is @xmath305 with probability 1 . hence the probability to exit through state @xmath305 is also @xmath302 .
( 85,25)(0,0 ) ( n0)(0,18)@xmath3 ( n1)(20,18)@xmath32 ( n2)(70,18)@xmath3 ( n3)(90,18)@xmath32 ( n4)(70,0)@xmath189 ( n5)(90,0)@xmath305 ( n1)@xmath306 ( n3)@xmath302 ( n4)@xmath306 ( n5)@xmath306 ( n0,n1)@xmath306 ( n2,n3)@xmath302 ( n2,n4)@xmath307 ( n3,n5)@xmath307 ( 85,25)(0,0 ) ( n0)(0,18)@xmath3 ( n1)(20,18)@xmath32 ( n2)(70,18)@xmath3 ( n3)(90,18)@xmath32 ( n4)(70,0)@xmath189 ( n5)(90,0)@xmath305 ( n4)@xmath306 ( n5)@xmath306 ( n0,n1)@xmath306 ( n1,n0)@xmath306 ( n2,n3)@xmath302 ( n3,n2)@xmath302 ( n2,n4)@xmath307 ( n3,n5)@xmath307 we now consider another example to illustrate further . consider the markov chain @xmath108 and @xmath303 in fig [ fig : illus_2 ] , where in @xmath108 it alternates between state @xmath3 and @xmath32 , and the discount factor is @xmath302 . if we start at state @xmath3 , the mean - discounted time at @xmath32 is given by @xmath308 the probability to exit through @xmath305 in @xmath303 in 2-steps is @xmath309 , in 4-steps is @xmath310 and so on .
hence the probability to exit through @xmath305 in @xmath303 is @xmath311 the above examples show how the mean - discounted time in @xmath108 and the exit probability in @xmath303 has the same value . _ ( of lemma [ lemm : poly ] ) .
_ we first write @xmath312 as follows : @xmath313 where @xmath314 , for all @xmath315 we have @xmath316 , @xmath317 , and @xmath318 for each @xmath319 . by the hypothesis of the lemma , for
all @xmath320 we have @xmath321 since every @xmath316 , multiplying the above inequalities by @xmath322 and summing over @xmath315 yields the desired result . _
( of lemma [ lemm_mc_multi ] ) .
_ we first observe that for a markov chain @xmath108 we have @xmath323 , i.e. , the value function for a state @xmath3 is obtained as the sum of the product of mean - discounted time of states and the rewards with @xmath3 as the starting state .
hence by lemma [ lemm : poly ] it follows that @xmath324 can be expressed as a ratio @xmath325 of two polynomials of degree at most @xmath199 over @xmath196 variables .
hence we have @xmath326 let @xmath327 . by definition for all
@xmath328 , if @xmath329 , then we have both @xmath330 and @xmath331 are between @xmath332 and @xmath333 .
it follows from lemma [ lemm : poly ] , with @xmath334 that @xmath335 thus we have @xmath336 hence we have @xmath337 we consider the case when @xmath338 , and the other case argument is symmetric .
we also assume without loss of generality that @xmath339 .
otherwise if @xmath340 , since rewards are non - negative , it follows that no state with positive reward is reachable from @xmath3 both in @xmath137 and @xmath138 ( because if they are reachable , then they are reachable with positive probability and then the value is positive ) , and hence @xmath341 and the result of the lemma follows trivially .
since we assume that @xmath338 and @xmath339 , we have @xmath342 & \leq & { { \mathsf{val}}}(g_2,{\mathsf{mdt}}({\vec{\lambda}},r))(s ) \cdot \big((1 + { \varepsilon})^{2\cdot|s|}-1\big ) \end{array}\ ] ] since the reward function is bounded by 1 , it follows that @xmath343 , and hence we have @xmath344 the desired result follows .
1 . for all player-1 mdps @xmath154\!]_\equiv}}$ ] , for all parity objectives @xmath1 and for all
@xmath22 we have @xmath345 & \leq & \displaystyle \bigg(1+\frac{{\mathit{dist}_a}(g_1,g_2)}{\eta}\bigg)^{2\cdot |s|}-1 \end{array}\ ] ] 2 . for @xmath7 ,
let @xmath246 . for all @xmath154\!]_\equiv}}$
] such that @xmath247 , for all parity objectives @xmath1 , every pure memoryless optimal strategy @xmath66 in @xmath137 is an @xmath6-optimal strategy in @xmath138 .
in other words , for the interval @xmath346 , every pure memoryless optimal strategy in @xmath137 is an @xmath6-optimal strategy in all structurally equivalent mdps of @xmath137 such that the distance lies in the interval @xmath346 . 1 . without loss of generality ,
let @xmath347 .
let @xmath66 be a pure memoryless optimal strategy in @xmath137 and such a strategy exists by theorem [ thrm_lit1 ] .
then we have the following inequality @xmath348 & \geq & { { \mathsf{val}}}(g_1 { \upharpoonright}\stra_1,\phi)(s ) - \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1\big ) \\[1ex ] & = & { { \mathsf{val}}}(g_1,\phi)(s ) - \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1\big ) \end{array}\ ] ] the ( in)equalities are obtained : the first inequality follows because the value in @xmath138 is at least the value in @xmath138 obtained by fixing a particular strategy ( in this case @xmath349 ; the second inequality is obtained by appying theorem [ thrm_mc_multi ] on the structurally equivalent markov chains @xmath350 and @xmath351 ; and the final equality follows since @xmath66 is an optimal strategy in @xmath137 .
the desired result follows .
2 . let @xmath154\!]_\equiv}}$ ] such that @xmath352 .
let @xmath66 be any pure memoryless optimal strategy in @xmath137 .
then we have the following inequality @xmath353 & = & { { \mathsf{val}}}(g_1,\phi)(s ) - \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1\big ) \\[1ex ] & \geq & { { \mathsf{val}}}(g_2,\phi)(s ) - 2\cdot \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1 \big ) .
\end{array}\ ] ] the first inequality is a consequence of theorem [ thrm_mc_multi ] applied on markov chains @xmath351 and @xmath354 ; the equality follows from the fact @xmath66 is an optimal strategy in @xmath137 ; and the infinal equality follows by applying the result of part 1 . hence to prove that @xmath66 is @xmath6-optimal in @xmath138 we need to show that @xmath355 we have @xmath356 the inequality follows from proposition [ prop_dist ] .
hence to prove inequality ( [ eq - diff1 ] ) it suffices to show that @xmath357 since @xmath246 , we obtain the desired inequality . _ ( of theorem [ thrm_tb_stochastic ] ) . _ the proof is essentially to repeat the proof of theorem [ thrm_mdp ] : as in mdps pure memoryless optimal strategies exist in turn - based stochastic games with parity objectives ( theorem [ thrm_lit1 ] ) ; and once a pure memoryless strategy is fixed in a turn - based stochastic game we obtain an mdp .
since theorem [ thrm_mdp ] extend the result of theorem [ thrm_mc_multi ] from markov chains to mdps , the proof for the desired result follows by mimicking the proof of theorem [ thrm_mdp ] and instead of using the result of theorem [ thrm_mc_multi ] for markov chains using the result of theorem [ thrm_mdp ] for mdps . _
( of lemma [ lemm_mdp_multi ] ) . _ the proof is essentially mimicking the proof of part(1 ) of theorem [ thrm_mdp ] . without loss of generality ,
let @xmath338 .
let @xmath66 be a pure memoryless optimal strategy in @xmath137 and such a strategy exists by theorem [ thrm_lit1 ] .
then we have the following inequality @xmath358 & \geq & { { \mathsf{val}}}(g_1 { \upharpoonright}\stra_1,{\mathsf{mdt}}({\vec{\lambda}},r))(s ) - \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1\big ) \\[1ex ] & = & { { \mathsf{val}}}(g_1,{\mathsf{mdt}}({\vec{\lambda}},r))(s ) - \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1\big ) \end{array}\ ] ] the ( in)equalities are obtained : the first inequality follows because the value in @xmath138 is at least the value in @xmath138 obtained by fixing a particular strategy ( in this case @xmath349 ; the second inequality is obtained by appying theorem [ thrm_mc_multi ] on the structurally equivalent markov chains @xmath350 and @xmath351 ; and the final equality follows since @xmath66 is an optimal strategy in @xmath137 .
the desired result follows . _
( of lemma [ lemm_conc_multi ] ) . _ the proof is essentially mimicking the proof of lemma [ lemm_mdp_multi ] . without loss of generality ,
let @xmath338 .
let @xmath66 be a randomized memoryless optimal strategy in @xmath137 and such a strategy exists by theorem [ thrm_lit1 ] .
then we have the following inequality @xmath358 & \geq & { { \mathsf{val}}}(g_1 { \upharpoonright}\stra_1,{\mathsf{mdt}}({\vec{\lambda}},r))(s ) - \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1\big ) \\[1ex ] & = & { { \mathsf{val}}}(g_1,{\mathsf{mdt}}({\vec{\lambda}},r))(s ) - \big((1 + { \mathit{dist}_r}(g_1,g_2))^{2\cdot|s|}-1\big ) \end{array}\ ] ] the argument for the inequalities are exactly the same as in lemma [ lemm_mdp_multi ] . the desired result follows .
we now show that the our quantitative bound for the value function difference is asymptotically optimal for small distances .
let us denote the absolute distance as @xmath6 , and quantitative bound we obtain in theorem [ thrm_conc_multi ] is @xmath359 , and if @xmath6 is small ( @xmath360 and @xmath6 close to zero ) , we obtain the following approximate bound @xmath361 we now illustrate with an example ( on structurally equivalent markov chains ) where the difference in the value function is @xmath262 , for small @xmath6 .
consider the markov chain defined on state space @xmath263 as follows : states @xmath164 and @xmath264 are absorbing ( states with self - loops of probability 1 ) and for a state @xmath265 we have @xmath362 i.e. , we have a markov chain defined on a line from @xmath268 to @xmath269 ( with @xmath268 and @xmath269 absorbing states ) and the chain moves towards @xmath268 with probability @xmath270 and towards @xmath269 with probability @xmath271 ( see fig [ figure : asym ] ) .
our goal is to estimate the probability to reach the state @xmath164 , and let @xmath272 denote the probability to reach @xmath164 from the starting state @xmath273
. then we have the following simple recurrence for @xmath265 @xmath363 and @xmath364 and @xmath365 .
we will consider @xmath366 such that @xmath6 is very small and hence higher order terms ( like @xmath367 ) can be ignored .
we claim that the values @xmath272 can be expressed as the following recurrence : @xmath368 , where @xmath369 .
the proof is by induction and is shown below : @xmath370 & = & ( \frac{1}{2}+{\varepsilon})\cdot v_{i-1 } + ( \frac{1}{2}-{\varepsilon})\cdot ( \frac{1}{2 } + { \varepsilon})\cdot c_i \cdot v_{i } \quad \text{(by inductive hypothesis $ v_{i+1}= ( \frac{1}{2}+{\varepsilon})\cdot c_i \cdot v_i$)}\\[1ex ] & = & ( \frac{1}{2}+{\varepsilon})\cdot v_{i-1 } + ( \frac{1}{4}-{\varepsilon}^2 ) \cdot
c_i \cdot v_{i}\\[1ex ] & = & ( \frac{1}{2}+{\varepsilon})\cdot v_{i-1 } + \frac{1}{4 } \cdot c_i \cdot v_{i } \qquad \text{(ignoring $ { \varepsilon}^2 $ ) } \end{array}\ ] ] it follows that @xmath371 .
hence we have @xmath372 & = & ( \frac{1}{2}+{\varepsilon})\cdot 1 + ( \frac{1}{2}-{\varepsilon})\cdot ( \frac{1}{2}+{\varepsilon})\cdot c_1 \cdot v_1 \\[1ex ] & = & ( \frac{1}{2}+{\varepsilon } ) + ( \frac{1}{4}-{\varepsilon}^2)\cdot c_1 \cdot v_1 \\[1ex ] & = & ( \frac{1}{2}+{\varepsilon } ) + \frac{1}{4}\cdot c_1 \cdot v_1 \qquad \text{(ignoring $ { \varepsilon}^2 $ ) } \end{array}\ ] ] thus we obtain that @xmath373
. then we have @xmath374 and then @xmath375 and so on .
finally we obtain @xmath376 as follows : @xmath377 .
observe that for the markov chain with @xmath274 , the states @xmath164 and @xmath264 are the recurrent states , and since the chain is symmetric from @xmath378 ( with @xmath274 ) the probability to reach @xmath264 and @xmath164 must be equal and hence is @xmath379 .
it follows that we must have @xmath380 .
hence we have that for @xmath7 , but very small , @xmath381 .
thus the difference with the value function when @xmath274 as compared to when @xmath7 but very small is @xmath382 . also observe that the markov chain obtained for @xmath274 and @xmath278 are structurally equivalent .
thus the desired result follows . | we consider two - player stochastic games played on a finite state space for an infinite number of rounds .
the games are _ concurrent _ : in each round , the two players ( player 1 and player 2 ) choose their moves independently and simultaneously ; the current state and the two moves determine a probability distribution over the successor states .
we also consider the important special case of turn - based stochastic games where players make moves in turns , rather than concurrently .
we study concurrent games with @xmath0-regular winning conditions specified as _ parity _ objectives .
the value for player 1 for a parity objective is the maximal probability with which the player can guarantee the satisfaction of the objective against all strategies of the opponent .
we study the problem of continuity and robustness of the value function in concurrent and turn - based stochastic parity games with respect to imprecision in the transition probabilities .
we present quantitative bounds on the difference of the value function ( in terms of the imprecision of the transition probabilities ) and show the value continuity for structurally equivalent concurrent games ( two games are structurally equivalent if the supports of the transition functions are the same and the probabilities differ ) .
we also show robustness of optimal strategies for structurally equivalent turn - based stochastic parity games .
finally , we show that the value continuity property breaks without the structural equivalence assumption ( even for markov chains ) and show that our quantitative bound is asymptotically optimal .
hence our results are tight ( the assumption is both necessary and sufficient ) and optimal ( our quantitative bound is asymptotically optimal ) . |
the recent discovery by novoselov et al @xcite that a single atomic layer of graphite usually known as graphene can be stable under some special experimental conditions opened widely the realm of a soon - to - come technology of graphene - based electronic devices .
the main transport mechanism of the charged carriers is in this case an enhanced quantum hall effect accompanied by a non - zero berry s geometric phase that has been experimentally observed @xcite . on the other hand , the quantum hall effect and the berry phase
have been also discussed for graphene bilayers @xcite that technologically could be even more interesting than graphene itself . motivated by these works , in this paper , we introduce the more general lewis - riesenfeld phases for the case of a graphene bilayer .
these phases are defined in a mathematical formalism called the ermakov - lewis approach , which is shortly presented .
for the bilayer , we will make usage of the recent gauged dirac model of jackiw and pi @xcite . with the aid of a mathematical procedure due to kravchenko @xcite we present the results in a series expansion in the bias potential @xmath0 applied to the bilayer graphene for any gauge potential @xmath1 .
we give detailed numerical results for a particular case of @xmath1 .
recently jackiw and pi introduced a simple gauged dirac model of bilayer graphene that can be reduced to the following coupled system of first - order differential equations @xcite @xmath2 where @xmath3 is a constant representing the external bias voltage and@xmath4 physically , function @xmath5 has two components : @xmath6 a scalar field @xmath7 that describes a particular dimerization called kekul distortion @xcite which can be also interpreted as characterizing the condensate arising from states bound by interlayer coulomb forces between particles in one layer and holes in the other ; the fingerprint of @xmath8 in @xmath5 is the parameter @xmath9 which should be an odd integer number .
@xmath6 a vector potential , @xmath1 , which was first introduced by jackiw and pi for a monolayer graphene @xcite to unpin the vortices and then studied in the context of bilayer graphene as well @xcite .
@xmath1 is a gauge field satisfying some appropriate conditions in the origin and at infinity . in @xcite
it was supposed that @xmath10 it is easy to see that the spinor components @xmath11 and @xmath12 are necessarily solutions of the following second - order differential equations obtained directly from system ( [ dir1])-([dir ] ) @xmath13 this couple of equations can be rewritten in the following way@xmath14 the latter form of the uncoupled system shows that the two functions @xmath11 and @xmath12 in the jackiw - pi model are supersymmetric partners in supersymmetric quantum mechanics and therefore many known supersymmetric results can be directly applied .
consider the sturm - liouville equation @xmath15 where @xmath16 and @xmath17 are complex - valued functions of a real variable @xmath18,\quad p\in c^{1}(a , b)$ ] is bounded and non - vanishing on @xmath19 $ ] and @xmath20 is an arbitrary complex constant .
suppose that there exists a solution @xmath21 of the equation @xmath22 on @xmath23 such that @xmath24 together with @xmath25 are bounded on @xmath19.$ ] kravchenko @xcite proved that the general solution of ( [ sch ] ) has the form@xmath26 where @xmath27 , @xmath28 are arbitrary complex constants and@xmath29 with @xmath30 and @xmath31 being defined by the following recursive relations@xmath32@xmath33{ll}% $ n\int_{a}^{r}\widetilde{x}^{(n-1)}(\xi)g_{0}^{2}(\xi)d\xi\qquad$ & $ \text{for an odd } n$\\ & \\ $ n\int_{a}^{r}\widetilde{x}^{(n-1)}(\xi)\frac{d\xi}{p(\xi)g_{0}^{2}(\xi ) } \qquad$ & $ \text{for an even } n$% \end{tabular } \right .
\label{xtil}%\]]@xmath34{ll}% $ n\int_{a}^{r}x^{(n-1)}(\xi)\frac{d\xi}{p(\xi)g_{0}^{2}(\xi)}\qquad$ & $ \text{for an odd } n$\\ & \\ $ n\int_{a}^{r}x^{(n-1)}(\xi)g_{0}^{2}(\xi)d\xi\qquad$ & $ \text{for an even } n.$% \end{tabular } \right .
\label{x}%\ ] ] this iterative scheme is appropriate for an easy implementation of numerical solutions and therefore for not exactly - solvable problems .
to apply kravchenko s procedure to the jackiw - pi uncoupled system we start with the equation for @xmath11 .
it is easy to find a particular solution @xmath35 from the factorized form of the @xmath11 equation ( [ second1 ] ) .
we immediately get @xmath36 and the general solution @xmath11 has the form @xmath37 , with @xmath38 where @xmath30 and @xmath31 are as in ( [ xtil ] ) and ( [ x ] ) , respectively , for @xmath39 .
the general solution @xmath40 of equation ( [ second2 ] ) can be obtained now by means of the darboux transformation @xmath41 @xmath42 by construction , @xmath43 and @xmath44 satisfy the following initial conditions @xmath45 with their wronskian being equal to @xmath46 consequently , the functions @xmath47 and @xmath48 take the initial values @xmath49
if one takes @xmath50 in ( [ second1 ] ) , the formulas ( [ u ] ) give @xmath51 where the expression in @xmath52 represents the well known construction of a second linearly independent solution of a schrdinger equation .
when @xmath53 , @xmath54 , the system ( [ dir1])-([dir ] ) reads as@xmath55 a solution @xmath21 in this case is an arbitrary constant , e.g. , @xmath56 . calculating @xmath30 and @xmath31 with the aid of ( [ xtil])-([x ] ) we obtain @xmath33{ll}% $ ( -1)^{\frac{n-1}{2}}r^{n},$ & $ \text{for an odd } n$\\ & \\ $ ( -1)^{\frac{n}{2}}r^{n},$ & $ \text{for an even } n$% \end{tabular } \right .
, \;x^{(n)}(r)=\left\ { \begin{tabular } [ c]{ll}% $ ( -1)^{\frac{n+1}{2}}r^{n},$ & $ \text{for an odd } n$\\ & \\ $ ( -1)^{\frac{n}{2}}r^{n},$ & $ \text{for an even } n.$% \end{tabular } \right.\ ] ] now substitution into ( [ u ] ) and ( [ v ] ) gives us the following solutions ( cf @xcite ) @xmath57 when @xmath58 and consequently @xmath59 , the superpotential @xmath5 takes the form @xmath60 . for system ( [ dir1 ] ) , ( [ dir ] ) with this particular potential , in @xcite the following solution is presented@xmath61 which is easily obtained from ( [ u ] ) and ( [ v ] ) choosing @xmath62 and calculating @xmath43 and @xmath63 indeed , for @xmath30 we obtain the formulas @xmath33{ll}% $ ( -1)^{\frac{n-1}{2}}\frac{n!!(k-1)!!}{(k+n)!!}r^{k+n},\qquad$ & $ \text{for an odd } n$\\ & \\ $ ( -1)^{\frac{n}{2}}\frac{(n-1)!!(k-1)!!}{(k+n-1)!!}r^{n},\qquad$ & $ \text{for an even } n$% \end{tabular } \right.\ ] ] leading us to the solution ( [ soljp ] ) multiplied by a constant .
we have now all the elements to proceed to specific calculations .
let us consider the dirac system ( [ dir1])-([dir ] ) choosing in ( [ fi ] ) the vector potential @xmath64 which is in agreement with conditions ( [ a ] ) .
the parameter @xmath65 plays the role of a screening length .
the superpotential ( [ fi ] ) then becomes @xmath66 and in this case the system ( [ schred1])-([schred2 ] ) has the following form@xmath67 the plots of @xmath11 and @xmath12 calculated within kravchenko s scheme are presented in fig.1 .
u_v_pi_2values.eps + fig.1 .
the solutions @xmath68 ( red ) and @xmath40 ( blue ) of the supersymmetric system ( [ scherd_sys])-([scherd_sys1 ] ) for @xmath69 and two cases of the arbitrary constants @xmath70 and @xmath71 : is for @xmath72 , and - - - is for @xmath73 .
we will now expound on the ermakov - lewis ( el ) formalism for the supersymmetric system ( [ schred1])-([schred2 ] ) . if the functions @xmath11 and @xmath12 are written in the form @xcite@xmath74 where @xmath75and @xmath76 are arbitrary constants , then both the amplitude functions @xmath77 and @xmath78 and the phase functions @xmath79 and @xmath80 are directly related to the ermakov - milne - pinney nonlinear equations @xcite @xmath81@xmath82 where @xmath83 is an arbitrary constant , in the following way . while @xmath77 and @xmath78 are solutions of ( [ ermakov ] ) with a respective potential , the phase functions @xmath79 and @xmath80 are expressed in terms of @xmath77 and @xmath78 as follows@xmath84 for @xmath77 and @xmath78 there is an elegant representation @xcite in terms of pairs of linearly independent solutions @xmath85 and @xmath86 of ( [ schred1 ] ) and ( [ schred2 ] ) respectively .
namely , general solutions of ( [ ermakov ] ) can be written in the following way @xmath87 where @xmath88 @xmath89 and @xmath90 are constants related by the equality @xmath91 , with @xmath92 being the wronskian @xmath93 ( which is a constant that does not change through the darboux transformation ) .
the solution of the ermakov equations allows us to obtain another important dynamic invariant quantity @xmath94 and @xmath95 respectively , called the lewis invariant @xcite .
@xmath94 is constructed in terms of any solution @xmath11 of ( [ schred1 ] ) and the corresponding solution @xmath77 of the ermakov - milne - pinney nonlinear equation ( [ ermakov ] ) ( and analogously @xmath95 is constructed in terms of @xmath12 and @xmath78 ) @xmath96
@xmath97 for the fixed interval @xmath19 $ ] the expressions @xmath98 give us the quantum number function depending on the energy @xmath99
@xcite,@xcite,@xcite @xmath100 used in standard quantum mechanics to identify bound states since at a bound state energy @xmath101 the function @xmath102 takes integer values @xmath103 and this does not depend on the choice of the constants @xmath104 , and @xmath90 .
however , if the energy is different from @xmath101 then ( [ qnumber ] ) is not uniquely determined , i.e. , for @xmath105 the function @xmath102 depends on the boundary conditions imposed to @xmath106 .
an important point of the el formalism is that the phase function ( [ phase ] ) also known as lewis - riesenfeld phase @xcite , has two components : the usual dynamic phase and a geometric phase that are given by the following formulas ( @xcite ) @xmath107 dr^{\prime}\label{dynamic}%\ ] ] and@xmath108 dr^{\prime } , \label{geom}%\ ] ] any of these phases can be calculated by employing kravchenko s iterative series ( [ u ] ) and ( [ v ] ) for the @xmath106 functions ( [ ro ] ) .
denoting@xmath109 we get @xmath110 ^{\frac{1}% { 2}}.\label{rou}%\ ] ] introducing the notations@xmath111 the amplitude function @xmath78 is calculated as follows@xmath112 ^{\frac{1}{2}}\label{rov}%\ ] ] the particular values of the functions @xmath77 and @xmath78 can be obtained by relating the constants @xmath104 and @xmath90 to ( [ condu1u2 ] ) and ( [ condv1v2 ] ) , that is@xmath113@xmath114 based on ( [ rou ] ) and ( [ rov ] ) , we have calculated the lewis - riesenfeld angular integrals ( [ dynamic ] ) and ( [ geom ] ) for the same decaying exponential profile of @xmath115.the results for the geometric lewis - riesenfeld phases are plotted in fig . 2 and fig . 3 of the paper for @xmath83 @xmath116 and @xmath117 , respectively , and for the same finite interval
. one can check by direct calculation that the values of the ermakov - lewis invariant ( [ inv ] ) do not change under tthe supersymmetric transformation , that is @xmath118 .
indeed , noticing that @xmath94 and @xmath95 are @xmath119-independent quantities , we can calculate them at one point only , e.g. , at @xmath120 and with arbitrary constants @xmath121 and @xmath122 .
we have @xmath123 taking into account that @xmath124 we obtain @xmath125 .
if the functions @xmath11 and @xmath12 are solutions of a coupled first - order system of the dirac type then the corresponding amplitude functions @xmath77 and @xmath78 are related through nonlinear dirac - ermakov relationships that in a particular case @xmath126 have been obtained previously by ioffe and korsch @xcite .
consider the equality@xmath127 substitution of @xmath128and @xmath129 leads to the relationship@xmath130 where @xmath131
. taking into account ( [ y1 ] ) , the functions @xmath132 and @xmath48 can be expressed as @xmath133 and @xmath134 . calculating the constants @xmath135 and
@xmath136 according to ( [ condv1v2 ] ) and ( [ condrov ] ) we get @xmath137 now taking the derivatives and substituting them into the expression for @xmath138 we obtain @xmath139 where one can notice that @xmath140 . employing the same procedure with respect to @xmath141 and taking into consideration that @xmath142 we arrive at the nonlinear dirac - ermakov system involving the amplitudes @xmath77 and @xmath78 and the phases @xmath143 of the dirac spinors @xmath11 and @xmath12@xmath144@xmath145 where again it is worth noticing that @xmath146 .
thus , there is a cross contribution to the squares of the intertwined amplitude functions of one - unit jumps in @xmath83 which is given by the squares of the derivatives of the corresponding eigenfunctions . for @xmath126 the system ( [ rou2])-([rov2 ] )
reduces to the result in @xcite .
to associate the spinor components @xmath11 and @xmath12 to the amplitude functions @xmath77 and @xmath78 the relationship between the constants @xmath70 and @xmath71 and @xmath104 and @xmath90 should be fixed .
for instance , choosing @xmath147 , @xmath148 we get @xmath149 , @xmath150 , @xmath151 .
the corresponding @xmath152 and @xmath153 for @xmath126 and @xmath117 are displayed in fig.2 , fig.3 . and fig.4 .
in addition , the amplitude functions for the same values of the parameters are shown in fig.5 .
notice that since the problem is undefined at the origin because of the @xmath154 singularity of @xmath155 the plots are displayed for the interval @xmath156 $ ] . for all plots we use the following three sets of values of the constants @xmath88 @xmath89 and @xmath90 : \1 .
@xmath157 @xmath158 @xmath159 .
these values imply @xmath160 @xmath161 @xmath162 .
@xmath163 @xmath164 @xmath165 . in this case
@xmath166 @xmath167 @xmath168 .
@xmath169 @xmath170 @xmath159 @xmath81leading to @xmath171 @xmath172 .
the values of @xmath173 and @xmath9 are chosen as follows : @xmath174 , @xmath175 and @xmath69 .
after testing kravchenko s numerical solutions in the context of jackiw and pi model of bilayer graphene we have developed an amplitude - phase approach of ermakov - lewis type for the same model and numerical solutions and for any radial profile of the gauge potential @xmath1 .
next , we have chosen a debye - screened form of the profile of @xmath1 to illustrate our results .
this choice could be of physical relevance when one considers the dynamic screening processes in the scattering of the charge carriers in the background of vortex configurations or other topological defects in the bilayer graphene .
our technique is general and therefore can be applied to any other physical problem of similar mathematical structure .
the geometric phase of lewis - riesenfeld type , which is more general than the berry phase , is calculated for the first time in the graphene context .
in addition , we provided a generalization of the ioffe - korsch nonlinear intertwining of the milne - pinney amplitude functions .
the first author would like to thank conacyt for a postdoctoral fellowship allowing her to work in ipicyt .
the second author wishes to thank conacyt for partial support through project 46980 .
the authors thank dr .
r. jackiw for useful communications . | in the context of bilayer graphene we use the simple gauge model of jackiw and pi to construct its numerical solutions in powers of the bias potential @xmath0 according to a general scheme due to kravchenko .
next , using this numerical solutions , we develop the ermakov - lewis approach for the same model .
this leads us to numerical calculations of the lewis - riesenfeld phases that could be of forthcoming experimental interest for bilayer graphene .
we also present a generalization of the ioffe - korsch nonlinear darboux transformation .
+ pacs numbers : 02.30.hq , 11.30.pb , 81.05.uw j. phys . a 42 ( 2009 ) 042004
( 11 pp ) + fast track communication |
the theoretical analysis of the masses and mixings of the light pseudoscalar and vector mesons has a long history , from the original gell - mann okubo mass formula @xcite through modern analyses based on chiral perturbation theory as summarized , for example , in @xcite .
the masses of the light pseudoscalar mesons are now discussed in terms of the breaking of the chiral @xmath2 symmetry that exists in qcd for massless quarks , with the octet mesons appearing initially as the eight goldstone bosons for the broken symmetry .
the su(3 ) singlet state @xmath3 and the vector mesons are treated separately . a perturbative treatment of the effect of quark masses on the pseudoscalar mesons , introduced through the quark mass matrix @xmath4 leads to the prediction that @xcite @xmath5 here @xmath6 is the average light - quark mass , @xmath7 , and @xmath8 and @xmath9 are the masses of the unmixed su(3 ) isospin-0 octet and singlet states @xmath10 and @xmath11 , @xmath12 is an unknown dynamical matrix element , and @xmath13 is the original mass of the @xmath14 , also unknown . finally , the original states @xmath10 and @xmath11 couple through an off diagonal matrix element @xmath15 to produce the observed mass eigenstates @xmath16(547 ) and @xmath17(958 ) @xcite .
the perturbative predictions are reasonably successful .
equation ( [ perturbative_masses ] ) contains the gell - mann okubo mass relation @xmath18 which is valid to about 20% when the @xmath19 is identified with the @xmath20 . the @xmath20-@xmath1 mixing angle defined by @xmath21
is predicted to be @xmath22 @xcite , at the edge of the range @xmath23 to @xmath24 obtained in various fits to decay data , for example , in @xcite .
the situation is different for the vector mesons , where the chiral symmetry arguments do not apply .
however , symmetry ideas from su(6 ) and quark - model and similar methods again lead to successful mass relations among the vector mesons , while dynamical ideas related to annihilation processes have been invoked to explain the different mixing patterns @xcite . in the present paper ,
i present a simple structural analysis of the mass and mixing problems for the nonets of pseudoscalar and vector mesons .
most of the ideas appear in the literature as applied to specific problems , and most of the results are also familiar .
however , by introducing some quark - mass effects which modify previous approaches and lead naturally to mass matrices of the most general type @xcite , i find a consistent overall picture in which , for example , the different mass and mixing patterns for the pseudoscalar and vector mesons follow from physical input , and the ratios of dynamical matrix elements in the two multiplets give information on the mixings .
i first treat the pseudoscalar nonet , with the @xmath1 included from the outset .
i then treat the vector nonet , and comment on the connections and differences , and conclude with a few overall remarks .
in this section , i discuss a generalized version of the mass matrix which allows actual fits to the masses of the pseudoscalar mesons . it will be useful to start in a `` quark model '' basis of effective meson fields @xmath26 and meson states @xmath27 , where @xmath28 label the flavor structure in a @xmath29 description of a flavor su(3 ) .
the general basis includes both the su(3 ) octet and singlet states .
the mass operator for the pseudoscalar mesons is then of the form @xmath30 i will work throughout with the matrix @xmath31 , and will first consider the neutral sector . the combinations @xmath32 , @xmath33 , and @xmath34 for the flavor labels @xmath28 are completely equivalent for massless quarks in qcd , so , with a suitable choice of phases , the initial mass matrix for the neutral states must have the same entry in every location corresponding to an su(3 ) singlet configuration , @xmath35 the factor @xmath12 in the second form has been extracted for convenience in writing later equations .
@xmath36 has eigenvalues zero in the su(3 ) octet states @xmath37 and @xmath10 , or any combination thereof , and the eigenvalue @xmath38 for the su(3 ) singlet @xmath11 .
the pion and @xmath19 are therefore massless , as expected for goldstone bosons .
the term @xmath39 contains contributions from the axial anomaly @xcite and from annihilation diagrams such as that in fig .
[ fig1](a ) .
both lead to the singlet structure in eq .
( [ m_0 ^ 2 ] ) when written in the @xmath0 basis .
the structure changes in the presence of quark masses , both through the addition of diagonal contributions proportional to the quark masses , and through the effect of those masses on the annihilation terms .
the effect of the latter is neglected in all discussions of which i am aware .
the modified mass matrix has an obvious structure @xmath40 in the @xmath0 representation , where the terms proportional to @xmath41 arise from mass insertions in the annihilation diagrams .
was apparently first suggested by isgur @xcite , but without the @xmath41 terms .
isgur identified what are here the @xmath42 terms with the annihilation diagrams , treated the diagonal terms phenomenologically as effective quark masses , and used the structure to explain the difference between the @xmath43 and @xmath44 mixings .
the general form of the mass matrix was given by morpurgo @xcite , but without the physical identification of the various contributions .
] i expect @xmath41 to be small , with @xmath45 .
a dispersion relation argument shows that the annihilation contribution to @xmath46 is positive in the symmetrical limit .
its magnitude is decreased by mass insertions in the loop , so @xmath41 should be negative and small .
acquires mass through the @xmath46 terms without any quark mass insertions . as a result
, the off - diagonal contributions proportional to @xmath47 can be complex , with @xmath48 still hermitian .
the imaginary part of @xmath41 should be small , and can be absorbed to first order in a change in the phase of @xmath11 .
i will therefore take @xmath41 as real . ]
the light quark masses are also small , @xmath49 . a transformation to the su(3 ) basis gives a considerably less transparent structure .
the resulting matrix can be simplified by dropping a diagonal term proportional to @xmath50 and off - diagonal terms which involve @xmath51 . both are small , with @xmath52 .
the off - diagonal terms induce a small mixing of the @xmath53 with the @xmath20 and @xmath1 , but first contribute to the physical masses in second order @xcite .
the reduced matrix is @xmath54 where @xmath6 and @xmath55 .
it is convenient to rewrite this as @xmath56 where @xmath57 is the 3@xmath583 unit matrix , @xmath59 is the 1@xmath582 null matrix , and @xmath60 is the @xmath612 matrix @xmath62 the matrix @xmath63 contributes a term @xmath64 to each of the eigenvalues of @xmath65 .
i will drop this term and work with the 2@xmath582 matrix @xmath60 which has eigenvalues @xmath66 and @xmath67 .
@xmath60 has the form of the general @xmath68 mass matrix @xcite , @xmath69 with @xmath70 real after a phase has been absorbed .
in particular , both involve three independent parameters .
i find it striking that the explicit inclusion of the annihilation terms already leads to this structure at first order in @xmath71 .
the kaon and charged pion masses have the usual simple structure which does not involve annihilation contributions , with @xmath72 the average kaon mass is @xmath73 , so @xmath74 the masses of the physical particles @xmath20 and @xmath1 are determined by the eigenvalues of @xmath60 , @xmath75 - \frac{1}{2}b_0\left[(3m_1 + 2(1+a)m_s)^2 - 8m_s(2m_1-a^2m_s)\right]^{1/2 } \nonumber\\ & \mbox{$\approx$ } & \frac{4}{3}b_0m_s\left[1 - \frac{2}{9}\left(1 + \frac{3}{2}a\right)^2 \frac{m_s}{m_1 } + \cdots\right ] , \\ \label{eta'_mass } \hat{m}_{\eta'}^2 & = & \frac{1}{2}b_0\left[3m_1 + 2(1+a)m_s\right ] + \frac{1}{2}b_0\left[(3m_1 + 2(1+a)m_s)^2 - 8m_s(2m_1-a^2m_s)\right]^{1/2 } \nonumber \\ & \mbox{$\approx$ } & 3b_0m_1+\frac{2}{3}b_0m_s\left[1 + 3a + \frac{4}{9}\left(1 + \frac{3}{2}a\right)^2 \frac{m_s}{m_1 } + \cdots \right ] , \end{aligned}\ ] ] where @xmath76 and @xmath77 .
the leading terms give the usual first - order perturbative result except for the independent appearance of @xmath41 in the expressions for @xmath78 and the @xmath43 mixing angle .
if @xmath41 is small , we obtain the inverse relations @xmath79 for @xmath80 and @xmath42 . the first becomes the gell - mann
okubo relation if @xmath80 is identified with @xmath81 as in eq .
( [ m_k ] ) .
these relations give the numerical estimates @xmath82 gev@xmath83 , @xmath84 gev@xmath83 , and @xmath85 for the physical masses @xmath86 gev@xmath83 , @xmath87 gev@xmath83 . the expansion parameter @xmath88 in eqs .
( [ eta_mass ] ) and ( [ eta_mass ] ) is therefore small , but not so small that the correction terms can be ignored .
since the `` first order '' fit to the observed masses automatically incorporates these corrections and other diagonal @xmath89 terms in the fitted values of @xmath46 and @xmath90 , it seems preferable to use the general form of the eigenvalues or @xmath60 in physical analyses .
the eigenvalues of the general mass matrix @xmath60 are determined by the conditions @xmath91\\ \label{det_relation } \hat{m}_{\eta'}^2\hat{m}_\eta^2 & = & { \rm det}\,\hat{m}_p^{\,2 } = b_0 ^ 2(4m_1m_s-2a^2m_s^2).\end{aligned}\ ] ] the @xmath92 term in the last expression is presumably small , and will be dropped in the immediately following equations for simplicity . in eq .
( [ a_limit ] ) farther from zero . ]
eliminating @xmath47 between the two equations and solving for @xmath46 gives the relation @xmath93^{1/2}.\ ] ] the argument of the radical is negative for the physical masses , and there is no real solution for @xmath46 , unless @xmath94/6\hat{m}_{\eta'}^2 \hat{m}_\eta^2=-0.084.\ ] ] this gives a useful constraint on our general parametrization of @xmath60 and requires @xmath41 to be negative , as expected . once a value of @xmath41 is given , eq .
( [ m1 ] ) determines @xmath42 , and @xmath80 then follows from eq .
( [ det_relation ] ) , @xmath95 with possible corrections of relative order @xmath96 .
we need one further condition to determine @xmath41 .
this is given within the @xmath97 sector by the mixing angle @xmath98 defined in eq .
( [ def : mixing_angle ] ) , @xmath99^{1/2}\right\}.\end{aligned}\ ] ] this angle is not well determined experimentally .
different analyses give values of @xmath23 to @xmath100 depending on the data used@xcite , and the theoretical connections assumed to hold between decay matrix elements and the meson masses and mixings .
a different approach is to take the value of @xmath101 as known .
the parameters @xmath46 and @xmath41 are then determined by the @xmath20 and @xmath1 masses , and @xmath98 can be predicted .
i will adopt this approach , but will allow for some difference between the effective values of @xmath80 in the kaon and @xmath43 systems by introducing a multiplicative correction factor @xmath102 , with @xmath103 .
this difference can arise from higher order effects in the chiral expansion , loop corrections , or the change in matrix elements expected in the change from a system with one heavy quark to systems which can have two .
all of these corrections can be encompassed in our general parametrization of @xmath60 ; their separation depends on dynamical calculation beyond the scope of this paper .
the net effect gives @xmath104 , with the expectation that @xmath105 is on the order of the fractional violation of the gell - mann
okubo relation .
the results obtained in our approach are summarized in figs .
[ fig2 ] and [ fig3 ] which show the fitted value of @xmath41 and the predicted mixing angle @xmath98 as functions of @xmath102 .
it appears from the results on mixing angles that @xmath102 is in the range @xmath106 , a reasonable result , with corresponding values of @xmath41 in the range @xmath107 .
the parameter @xmath41 is therefore small , as expected .
an argument given later in connection with eq .
( [ b0/b0 ] ) suggests that , when the vector mesons are included in the analysis , values @xmath108 , @xmath109 , and @xmath110 are favored .
this predicted mixing is in the center of the ranges found in various analyses @xcite .
the situation for the vector mesons @xmath111 , @xmath112 , and @xmath113 is quite different . in the @xmath0 basis , the initial mass matrix has the form of the pseudoscalar mass matrix in eq .
( [ m^2_complex ] ) with the addition of a diagonal term @xmath114 which gives a common mass to all the vector states .
no such term was not allowed in the pseudoscalar sector by the requirement that the pion and @xmath19 appear initially as goldstone bosons of the broken @xmath115 symmetry .
furthermore , the elements analogous to the terms @xmath42 in eqs .
( [ m_0 ^ 2 ] ) or ( [ m^2_complex ] ) have no anomalous component , so obtain contributions only from annihilation diagrams .
the leading vector annihilation diagram involves three - gluon exchange as shown in fig .
[ fig1](b ) . the corresponding matrix element , which i will denote by @xmath116 , is expected to be much smaller than the two - gluon contribution to the pseudoscalar masses from fig .
[ fig1](a ) , but still positive with @xmath117 .
the relation @xmath118 from the pseudoscalar sector then implies that @xmath119 .
the effect of mass insertions on the already - small @xmath120 terms should also be small , with @xmath121 .
dropping those corrections , we obtain the simplified vector mass matrix @xmath122 the matrix elements @xmath80 and @xmath123 differ in the dynamical quark model only by the effects of spin - dependent interactions on the wave functions and are expected to be similar in magnitude , @xmath124 .
in contrast to @xmath48 , which was dominated by the flavor - singlet structure of the @xmath46 terms in eq .
( [ m_0 ^ 2 ] ) with @xmath125 appearing as a perturbation , @xmath126 is dominated by the the strange - quark mass term .
there is consequently no reason to transform to the su(3 ) basis used to pick out the singlet structure .
the @xmath0 basis is natural .
it is useful , however , to transform to a flavor su(2 ) basis @xmath127 , @xmath128 for the light - quark sector .
this isolates the isovector component @xmath129 , and brings the mass matrix to the form @xmath130 where i have dropped off - diagonal terms in @xmath51 which contribute linearly to the @xmath131 mixing @xcite , but only quadratically to the masses .
the matrix @xmath132 is @xmath133 where @xmath55 as before .
the mass of the @xmath111 meson is clearly given by @xmath134 .
the eigenvalues of @xmath132 , taken to first order in the small quantity @xmath135 , are @xmath136 these relations give @xmath137 gev@xmath83 , @xmath138 gev@xmath83 , and @xmath139 gev@xmath83 .
thus , @xmath140 , and @xmath119 as expected .
the approximate degeneracy of the @xmath111 and @xmath112 masses follows from the smallness of @xmath120 .
this is expected in qcd as argued above .
there is no constraint on the extra parameter @xmath141 in the general case since @xmath142 .
the average @xmath143 mass squared @xmath144 is defined by @xmath145 the combination of the two preceding equations gives the sum rule @xmath146 this is satisfied to 6% ( @xmath147 ) . however , this sum rule is subject to some uncertainty theoretically because of different effect of the o(@xmath89 ) contributions and loop corrections on the @xmath44 and @xmath143 systems .
if the uncertainties are incorporated in a scaling of the @xmath143 contribution with @xmath148 , the scale factor @xmath149 is 1.06 , in the favored range for the scale factor @xmath102 for the pseudoscalar mesons . a comparison of eqs .
( [ estimates ] ) and ( [ vmasses ] ) leads to the rough estimate @xmath150 while the ratio @xmath151 gives @xmath152 . with either estimate , @xmath153 as expected .
if i include the leading correction to @xmath154 from eq .
( [ eta_mass ] ) , i find the more precise relation @xmath155^{-1}.\ ] ] this is satisfied for @xmath156 , @xmath157 .
the value of @xmath102 is in the range @xmath158 consistent with analyses of @xmath43 mixing , and gives the value @xmath159 .
the whole picture is therefore consistent .
i will define the mixing angle @xmath160 connecting the isospin zero state @xmath161 with @xmath162 so that @xmath163 the results above give @xmath164 numerically , @xmath165 , so there is very little mixing .
the angle @xmath166 defined with respect to the su(3 ) basis @xmath167 , @xmath168 @xcite is related to @xmath160 by @xmath169 where @xmath170 is the angle for so - called `` ideal '' su(3 ) mixing .
the sum gives @xmath171 , in good agreement with experiment .
however , i find little reason to use the su(3 ) description . the inequality @xmath172 picks out the `` @xmath0 '' basis states @xmath173 as giving the natural description of the problem , with the @xmath111 split off in the isospin triplet substate .
the physical masses then correspond yield a small mixing angle @xmath174 without special assumptions .
it is interesting to compare this result to that for the pseudoscalars . in the latter case , @xmath175
, so the su(3 ) singlet is not clearly singled out , and it is less certain what basis is to be preferred .
@xmath60 is given in the @xmath161 , @xmath162 basis by @xmath176 the @xmath177 mixing angle @xmath178 defined by the relations @xmath179 is given by @xmath180^{1/2}\right\ } \\
& \approx & \sqrt{2}\left(m_1+am_s\right)/\left[m_1 + 2(1+a)m_s\right].\end{aligned}\ ] ] this gives a rather large mixing angle in the range 47@xmath181 to 33@xmath181 for @xmath102 from 0.9 to 1.2 , a consequence of the similar sizes of @xmath46 and @xmath47 .
the su(3 ) basis is therefore somewhat better physically .
in contrast , given the result @xmath175 for the pseudoscalars , there is a clear physical argument that @xmath182 , so that the @xmath177 basis is preferred for the vectors , with the mixing small corresponding to `` ideal '' mixing in the su(3 ) basis .
my conclusions are simple : there is a substantial advantage in treating the nonets of pseudoscalar and vector meson together in analyzing the patterns of masses and mixings .
it is important , in particular , to include the @xmath183 in the analysis to get a consistent overall picture .
most of the results presented here are well - known , though not treated as a whole in other discussions .
however , the complete analysis given here introduces a new feature through the through the inclusion of the leading quark mass corrections to the annihilation diagrams in fig .
this leads to the general mass matrix for the pseudoscalar mesons in eq .
( [ m^2_complex ] ) , and a similar result for the vector mesons .
these reduce to the simpler forms in eqs .
( [ m=1+mhat ] ) and ( [ m_v+hatm_v ] ) if small off - diagonal terms proportional to @xmath51 are neglected .
the residual 2@xmath582 mass matrices have a completely general form .
i show in the case of the pseudoscalar mesons that the physical @xmath20 and @xmath1 masses can only be fitted if the annihilation parameter @xmath41 in the mass matrix is nonzero and negative .
the wide splitting of the @xmath184 , @xmath20 , and @xmath1 masses , and fairly large mixing angle @xmath98 for the pseudoscalars , reflect the fact that the strange - quark mass insertion @xmath80 and the basic su(3 ) singlet term @xmath42 are similar in magnitude .
the situation is much simpler for the vector mesons . while the initial mass matrix has the same formal structure as that for the pseudoscalar mesons , the large anomalous contribution to the singlet component of the latter is missing , and the annihilation diagram is suppressed by the necessity of having at least three gluons in the intermediate state . as a result , the parameter @xmath120 analogous to @xmath46
is expected to be small in qcd , with @xmath185 , and the mass matrix simplifies .
the smallness of @xmath120 leads to the near - degeneracy of the @xmath111 and @xmath112 masses , and to the small mixing of the @xmath112 and @xmath113 mesons in the natural @xmath0 basis .
finally , i show that the ratio of the matrix elements @xmath12 and @xmath186 for the pseudoscalar and vector mesons is close to one , as expected . by calculating the ratio two ways , i find that overall consistency is obtained for @xmath109 and scale factors @xmath187 .
the latter parametrize the higher order mass and loop corrections which differ for the @xmath188 and the @xmath43 or @xmath44 sectors .
the @xmath43 mixing angle in the usual su(3 ) basis is then predicted to be @xmath159 .
the @xmath44 mixing is small in the natural @xmath0 basis , @xmath189 , so is nearly `` ideal '' in an su(3 ) basis .
m. gell - mann , phys .
125 * , 1067 ( 1962 ) .
s. okubo , prog .
* 27 * , 949 ( 1962 ) .
j. gasser and h. leutwyler , phys .
rep . * 87 * , 77 ( 1982 ) .
j. f. donoghue , e. golowich , and b. r. holstein , _ dynamics of the standard model _ , ( cambridge university press , cambridge , england , 1992 ) , chaps .
vii and xiii .
particle data group , d. e. grom _
et al . _ ,
j. c * 15 * , 1 ( 2000 ) , sec .
j. f. donoghue , b. r. holstein , and y .- c .
r. lin , phys .
lett . * 55 * , 2766 ( 1985 ) .
f. j. gilman and r. kauffman , phys .
d * 36 * , 2761 ( 1987 ) ; * 37 * , 3348(e ) ( 1988 ) .
a. bramon and m. d. scadron , phys
. lett .
b * 234 * , 346 ( 1990 ) .
a. bramon , r. escribano , and m. d. scadron , eur .
j. c * 7 * , 271 ( 1999 ) .
e. p. venugopal and b. r. holstein , phys .
d * 57*,4397 ( 1998 ) .
n. isgur , phys .
d * 13 * , 122 ( 1976 ) .
g. morpurgo , phys .
d * 41 * , 2865 , ( 1990 ) . | i present a simple discussion of the masses and mixings of the light pseudoscalar and vector mesons based on a `` @xmath0 '' description of the effective field theory .
the analysis includes @xmath1(958 ) from the beginning , and is largely concerned with structural questions . while the final results are mostly known , the method gives insight into the general form of the meson mass matrices and the different character of the mass splittings and mixings in the pseudoscalar and vector multiplets , and provides a coherent overall picture |
[ 1 ] w. a. bardeen , i. bars , a. j. hanson and r. d. peccei , `` a study of the longitudinal kink modes of the string , '' phys
. rev .
d * 13 * , 2364 ( 1976 ) .
`` quantum poincar covariance of the d = 2 string , '' phys .
d * 14 * , 2193 ( 1976 ) .
l. mezincescu and p. k. townsend , `` anyons from strings , '' phys .
lett . * 105 * , 191601 ( 2010 ) [ arxiv:1008.2334 [ hep - th ] ] ; `` quantum 3d superstrings , '' phys .
d * 84 * , 106006 ( 2011 ) [ arxiv:1106.1374 [ hep - th ] ] .
m. b. green and j. h. schwarz , `` covariant description of superstrings , '' phys .
b * 136 * , 367 ( 1984 ) ; `` properties of the covariant formulation of superstring theories , '' nucl . phys .
b * 243 * , 285 ( 1984 ) .
l. brink , p. di vecchia , p. s. howe , `` a locally supersymmetric and reparametrization invariant action for the spinning string , '' phys .
* b65 * ( 1976 ) 471 - 474 .
r. jackiw and v. p. nair , `` relativistic wave equations for anyons , '' phys .
d * 43 * , 1933 ( 1991 ) .
m. s. plyushchay , `` relativistic particle with torsion , majorana equation and fractional spin , '' phys .
b * 262 * , 71 ( 1991 ) .
j. f. arvis , `` the exact q anti - q potential in nambu string theory '' phys .
b * 127 * ( 1983 ) 106 .
e. bergshoeff and p. k. townsend , `` super d - branes , '' nucl .
b * 490 * ( 1997 ) 145 [ arxiv : hep - th/9611173 ] . | we explain how fractional spin and statistics are relevant to ( super)strings in a three - dimensional ( 3d ) minkowski spacetime .
umtg-26 + damtp-2011 - 103 -1truecm [ [ sec:1 ] ] the nambu - goto ( ng ) string , in a d - dimensional minkowski spacetime , can be quantized in a unitary gauge that guarantees the absence of ghosts , but this can only be done by breaking manifest lorentz invariance , in which case lorentz invariance of the quantum theory is not guaranteed .
in fact , as shown long ago by goddard et al .
@xcite , who quantized the ng string in the light - cone gauge , the quantum theory breaks lorentz invariance unless @xmath0 .
analogous calculations for superstrings lead instead to @xmath1 as the critical dimension .
a ghost - free quantization below the critical dimension is possible but only after the introduction of an additional liouville field on the worldsheet @xcite or other `` longitudinal '' variables @xcite that are absent in the critical dimensional .
however , there is an exceptional subcritical dimension for which a unitary gauge quantization preserves lorentz invariance without the need to introduce any additional variables : @xmath2 .
the usual difficulties with the closure of the lorentz algebra are trivially absent for @xmath2 @xcite and a detailed study of closed oriented `` 3d '' strings @xcite shows not only that are there no lorentz anomalies but also that there are no supersymmetry anomalies of the 3d green - schwarz ( gs ) superstrings .
we recall that the classical gs superstring action , which exists for @xmath3 @xcite , is an extension of the ng action to describe a string moving in an @xmath4 or @xmath5 superspace .
quantization in the critical dimension ( @xmath1 ) leads to a supersymmetric spectrum of states , with massless ground states , and the same is true for @xmath2 .
we shall review the main results to date of light - cone gauge quantization of 3d strings , focusing on closed strings but with some comments on open strings .
we begin with a brief review of the hamiltonian formulation of the ng string .
the canonical variables are maps from the string worldsheet , parametrized by @xmath6 to a minkowski phase space , with @xmath7-vector coordinates @xmath8 .
the action is @xmath9 = \int \ !
d\tau \int_0^{2\pi } \ !
\frac{d\sigma}{2\pi } \left\{\dot { \bx } \cdot { \bp } - \frac{1}{2}\ell \left[{\bp}^2 + ( t\bx^\prime)^2\right ] - u\ , \bx^\prime \cdot \bp \right\}\ , , \ ] ] where the coordinate @xmath10 of the string takes values in the interval @xmath11 $ ] .
if we insist on periodicity in @xmath10 then we have the action for the closed string ; otherwise we have an open string .
elimination of the momenta @xmath12 ( which can be achieved in the path - integral by gaussian integration ) leads to the `` polyakov '' action @xcite : @xmath13 \rightarrow s[\bx;g ] = -\frac{t}{4\pi } \int \ !
d^2\xi \
, \sqrt{-\det g}\ , g^{ij } \partial_i \bx \cdot \partial_j\bx\ , , \ ] ] where @xmath14 , and the independent worldsheet metric @xmath15 is conformally equivalent to the metric @xmath16 any conformal factor drops out of the action due to conformal invariance .
it reappears quantum - mechanically in polyakov s covariant path - integral quantization @xcite but is not needed for a light - cone gauge quantization in the critical dimension @xcite and the same is true for @xmath2 , at least for free strings .
the lagrange multipliers @xmath17 appearing in the action ( [ hamactbos ] ) are analogous to the lapse and shift variables of the hamiltonian formulation of general relativity . as in that case
, the constraint functions generate gauge invariances that are equivalent to reparametrization invariance , in this case of the string worldsheet .
in addition to its gauge invariances , the action ( [ hamactbos ] ) is poincar invariant for periodic ( closed string ) or free - end ( open string ) boundary conditions . in 3d
the poincar noether charges are @xmath18 the two casimirs of the 3d poincar group are @xmath19 and these may be used to classify the unitary irreducible representations @xcite , in particular @xmath20 is the mass . the quantity @xmath21 is sometimes called the `` spin '' but we prefer to reserve this term for the absolute value @xmath22 and call @xmath21 the `` relativistic helicity '' , which we usually abbreviate to `` helicity '' .
so defined , spin is unchanged by parity , whereas the helicity changes sign . as both the ng string and
its gs superstring extensions preserve parity , we should expect all massive states of non - zero spin to appear in parity doublets of opposite sign helicity , and it can be shown that this is indeed the case @xcite . in our discussion of 3d gs superstrings we shall focus on the @xmath5 case for which the superspace coordinates are @xmath23 , where @xmath24 are a pair ( @xmath25 ) of two - component anti - commuting majorana spinors .
the action is obtained in two steps .
the first is to replace the differentials @xmath26 by the super - translation invariant 1-forms @xmath27 , where @xmath28 are the 3d @xmath29 dirac matrices and summation over the index @xmath30 is implicit .
the next step is the addition of a wz term constructed from a closed super - poincar invariant 3-form @xmath31 . for @xmath2
there is a 1-parameter family of such 3-forms , which we may parametrize in terms of a unit @xmath32-vector @xmath33 : @xmath34 the gs superstring is obtained by choosing @xmath35 , and a coefficient for the resulting wz term that leads to an additional fermionic gauge invariance .
after gauge - fixing , this leads to a cancellation of zero - point energies and hence massless ground states . in the quantum theory
each oscillation mode of a string corresponds to some polarization state of a particle of definite mass - squared .
the oscillator ground - state energy of the ng string is negative in the critical dimension , corresponding to a tachyon .
it is undetermined in 3d and this complicates a discussion of the spectrum .
for this reason , we focus here on the gs superstrings for which the ground states are those of a massless superparticle .
for the critical dimension this leads to states corresponding to the polarization modes of a @xmath1 linearized supergravity theory .
for @xmath2 it again leads to an irreducible massless supermultiplet .
for @xmath5 it consists of two bosons and two fermions , which could be identified as the states of a massless vector supermultiplet ; this option would introduce a vector that could couple to objects carrying the central charge allowed by the @xmath5 supersymmetry algebra .
we say `` option '' because spin is not defined for massless particles in 3d , despite the fact that there is still a distinction between bosons and fermions @xcite .
all excited states of the 3d gs superstrings are massive , so let us now look at the structure of massive 3d supermultiplets .
we focus on the @xmath5 case , for which there are two hermitian spinor charges @xmath36 , @xmath25 ; the superfix @xmath37 indicates the spinor component . within a given representation
, we may define the new non - hermitian supercharges @xmath38 \ , , \qquad { \cal p}_-\equiv \frac{1}{\sqrt{2 } } \left({\cal p}_1 - { \cal p}_0\right).\ ] ] the coefficients are particular to the conventions used in @xcite , and in these conventions the anticommutation relations of the @xmath39 supercharges become equivalent to the anticommutation relations @xmath40 where we have allowed for a central charge @xmath41 .
it can be shown that the charges @xmath42 lower the helicity @xmath21 by @xmath43 , and that their hermitian conjugates @xmath44 raise it by @xmath43 . associated with any state annihilated by the @xmath42
we then get a supermultiplet by successive actions of the operators @xmath44 on this state .
as we explain at the conclusion of this contribution , the possibility of a central charge is relevant to non - perturbative states in the spectrum , assuming that there is an interacting string theory .
however , no state in the spectrum of a free @xmath5 superstring carries a central charge , so we now set @xmath45 . in this case , the construction summarised above leads to four - state supermultiplets with helicities @xmath46 for some average helicity @xmath47 , which is the superhelicity ( the value , divided by @xmath20 , of the casimir of the super - poincar group that generalizes @xmath48 ) . because parity flips the sign of superhelicity , the unique parity - preserving _
irreducible _ supermultiplet ( with @xmath45 ) has @xmath49 , and hence felicities @xmath50 .
this is just the scalar supermultiplet of 4d @xmath4 supersymmetry reduced to 3d .
one can show that all massive states of the string have at least a 4-fold degeneracy , in addition to the 4-fold degeneracy implied by supersymmetry , so the number of states at each mass level is a multiple of 16 . at the first level
there are precisely 16 states which appear in 4 copies of the scalar supermultiplet just described . at the next level
there are 8 copies of this supermultiplet plus 4 copies of the spin-2 supermultiplet with @xmath51 and 4 copies of the parity - conjugate spin-2 multiplet with @xmath52 .
up to this point , there is nothing really unexpected . at the third level
we get another 8 copies of the scalar supermultiplet with @xmath49 but the remaining 28 + 28 supermultiplets , which appear in parity doublets of opposite sign superhelicity , all have _ irrational _
superspin @xmath53 , and hence all states in these supermultiplets have irrational spin .
although the details differ , irrational spins also appear in the spectrum for @xmath4 and for the 3d ng string .
irrational spins are possible because of a special feature of 3d : the universal cover of the lorentz group is not its double cover , the spin group @xmath54 ; if it were then @xmath21 would be quantized such that @xmath55 , as in 4d . instead
, the universal cover is an _
infinite _ cover with representations for _ any _ value of @xmath21 .
in particular , in any theory with irrational spins the `` lorentz '' group must be the universal cover of @xmath56 .
this has significant implications for quantum statistics via the 3d spin - statistics theorem ( see e.g. @xcite ) which relates the statistical phase @xmath57 to the helicity @xmath21 via the formula @xmath58 so a particle of spin @xmath21 such that @xmath59 is neither a boson nor a fermion but an _
anyon_. actually , it would be preferable to interpret `` anyon '' to mean @xmath60 because those cases for which @xmath61 for integer @xmath62 are also very special ; these are `` semions '' ( see e.g. @xcite )
. the presence of anyons in the spectrum of 3d strings may explain why @xmath2 does not appear to be exceptional when covariant quantization methods are used , e.g. the polyakov path - integral approach .
these methods may tacitly assume that the `` lorentz '' group is @xmath54 .
it is worth noting here that covariant quantization is already problematic for a 3d _ particle _ if it is an anyon , so it is not surprising that standard approaches to the covariant quantization of strings appear to exclude @xmath2 .
one difficulty can be seen from the fact that manifestly lorentz covariant anyonic wave equations are necessarily infinite - dimensional @xcite .
we conclude with a few comments on open 3d strings . in this case
each endpoint may be free or fixed , or it may be fixed to lie on some curve .
a unitary gauge quantization of the ng string with fixed ends was carried out by arvis @xcite , who concluded that rotational invariance is broken unless @xmath0 .
however , @xmath2 is again exceptional because the relevant rotation group is then trivial , or abelian if the endpoints coincide . in the case of open gs superstrings ,
the boundary conditions must break some of the supersymmetry but there are cases in which half is preserved .
let us focus on the @xmath5 superstring . for free endpoints ,
the @xmath5 supersymmetry is broken to @xmath4 , as in the critical dimension .
if the endpoints are fixed to lie on some curves , which may be interpreted as d - strings , then these curves must be parallel straight lines to preserve half the supersymmetry .
this possibility is expected in view of the ambiguity noted earlier in the superstring action ; we may associate the choice @xmath63 in ( [ wz ] ) with the d - string , by analogy to critical - dimension iib superstrings ( see e.g. @xcite ) . the remaining case in which half
the supersymmetry is preserved occurs when both ends are fixed .
if they are fixed at the same point then rotational invariance in the plane is preserved , and this becomes the so(2 ) symmetry of a standard supersymmetric quantum mechanics with two supersymmetry charges .
we do not know if 3d strings can interact , but if they can it would be natural to interpret the endpoints as d0-branes , which would be described by a centrally - charged @xmath5 superparticle with a mass @xmath20 saturating the bogomolnyi bound @xmath64 that follows from the algebra ( [ sqm ] ) .
quantization of this superparticle action leads to a @xmath5 supermultiplet in which all states have spin @xmath65 @xcite ; in other words , the would - be d0-branes are semions
. if the effective action for a pair of such d0-branes is , in the non - relativistic limit , a supersymmetric @xmath66 matrix mechanics with two supersymmetries , as is suggested by critical superstring theory , then the spin-@xmath65 nature of the d0-branes should also be apparent from their statistical phase .
indeed , this is precisely the conclusion of pedder , sonner and tong , who considered this model and computed the statistical phase as a berry phase @xcite .
we are grateful to michael green , julian sonner and david tong for helpful discussions .
lm acknowleges partial support from national science foundation award 0855386 . |
nucleon transfer reactions induced by hadronic probes at intermediate energies favour the excitation of high angular momentum states lying above particle emission threshold @xcite .
the observed structures originate from the coupling of the initial single - particle mode with more complex states .
the coupling of single - particle states with surface vibrations is mainly responsible for the damping of the single - particle mode @xcite .
the particle decay of highly excited states gives the opportunity to study in detail the damping process .
for example , the relative contributions of the direct and statistical components to the damping of single - particle mode can be found . up to now
, experimental data and theoretical calculations are available mainly for the neutron decay of high - lying single - particle modes .
very recently , the proton decays of high - lying states in @xmath0sc , @xmath1cu and @xmath2nb have been measured @xcite .
in odd nuclei , the simplest excited states can be described as admixtures of single - particle states or quasiparticle states coupled to collective excitations ( or phonons ) of the even - even core .
this weak coupling picture has been successfully applied to obtain the strength functions of a variety of odd nuclei @xcite .
a method for calculating particle escape widths in the framework of the quasiparticle - phonon model ( qpm)@xcite has been suggested in ref .
@xcite where the inclusive semi - direct neutron decays of high - lying states in @xmath3pb have been studied using the general procedure of refs . @xcite which was established for the case of nucleon emission from giant resonances . in a previous work @xcite
we have extended the method of ref.@xcite in order to calculate non - statistical particle decays of excited states in odd nuclei leading to exclusive channels which correspond to the ground and low - lying excited states of the even - even core . in the present work
we apply our method to study the proton decay . using the qpm
we calculate the partial cross sections and branching ratios for the proton decay of the high angular momentum states in @xmath0sc , @xmath1cu and @xmath2nb and compare them with experimental data .
this paper is organized as follows : in sec .
ii we describe briefly our theoretical approach to treat the direct nucleon decay of high angular momentum states of single - particle type . in sec .
iii a comparison of the calculated and measured branching ratios for non - statistical proton decay of high - lying states in @xmath0sc , @xmath1cu and @xmath2nb is presented . finally , in sec .
iv conclusions are drawn .
the projection operator method is a convenient approach to treat problems involving single - particle continua in a many - body context .
the general formalism was introduced by yoshida and adachi@xcite and it has been applied to studies of high - lying states in odd nuclei@xcite .
the detailed expressions can be found in ref.@xcite . here , we simply recall the main features .
we write the hamiltonian of the a+1 system in the form : @xmath4 the first term describes the motion of a particle in an average potential @xmath5 created by the particles in the core : @xmath6 the core hamiltonian is a sum of single - particle hamiltonians @xmath7 and two - body residual interactions @xmath8 : @xmath9 the last term of @xmath10 is a sum of interactions between the odd particle and the core particles : @xmath11 the physical spectrum of @xmath12 consists of a small number of bound states @xmath13 and a continuum of scattering states @xmath14 which form altogether a complete , orthogonal basis .
the projection operator method consists in introducing another complete set of orthogonal basis states which is a direct sum of two complementary subsets , a first subset of discrete , orthonormal states @xmath15 which span the single - particle @xmath16 space and a second subset of @xmath17 continuum states spanning the complementary @xmath18 space
. we will denote by @xmath19 ( @xmath20 ) the creation and annihilation operators of state @xmath21 ( @xmath22 ) . in the spirit of the qpm@xcite
the hamiltonian @xmath23 is treated in the random - phase approximation ( rpa ) in a discrete space , i.e. , the particle - hole configurations of rpa are built only with q - space states .
we denote by @xmath24 and @xmath25 the energies and creation operators of these rpa states which describe core excitations .
if @xmath26 represents the rpa ground state of the core , the properties of the ( a+1)-nucleus except for its nucleon decay properties can be described in terms of the one - particle states @xmath27 and one - particle - plus - phonon states @xmath28 \mid 0 > $ ] .
we can write : @xmath29\big ) \mid 0 > ~. \label{f5}\end{aligned}\ ] ] we call q space the space spanned by the ( real ) state vectors @xmath30 and @xmath31 the corresponding projection operator
. the amplitudes @xmath32 and @xmath33 , and the energies @xmath34 of @xmath30 are determined by diagonalizing @xmath10 in the rpa , i.e. , one solves : @xmath35 = \omega_i d^\dagger_i~ , \label{f6}\end{aligned}\ ] ] within the approximation of commutator linearization .
the distribution of @xmath36 represents the strength function from which one can deduce the spectroscopic factors@xcite . to allow for nucleons to decay
it is necessary to introduce state vectors where the odd particle has a non - zero probability of being at infinity .
this is achieved by constructing the p space complementary to q space and consisting of all states which are linear combinations of the following one - particle and one - particle - plus - phonon configurations : @xmath37 \mid 0 > ~. \label{f7}\end{aligned}\ ] ] the present definition of p space neglects continuum effects on the phonons @xmath25 which can also in principle couple to the continuum and emit nucleons by themselves . actually , these effects should have a small influence on the non - statistical particle decay of the ( a+1)-nucleus since the most important phonons contributing to the particle - phonon coupling are the low - lying collective states of the core .
the direct sum of q space and p space is by construction the complete particle - plus - phonon space in which the hamiltonian @xmath10 should be solved .
it is completely equivalent to solve in the simpler q space the more complicated effective hamiltonian : @xmath38 where @xmath39 is the projection operator onto p space ( @xmath40 ) and @xmath41 is the energy of the system .
the hamiltonian @xmath42 is complex and energy dependent . for each value of @xmath41 one
has to find the set of complex states and eigenenergies : @xmath43 satisfying : @xmath44 = \omega_i { \cal d}^\dagger_i~. \label{f10}\end{aligned}\ ] ] we consider a direct transfer reaction @xmath45 followed by a sequential decay @xmath46 where the ( a+1)-nucleus in a highly excited state decays by a semi - direct proton emission . in this process
, the @xmath47 states will act as doorway states .
if we describe the reaction mechanism in a simple approach , e.g. , a distorted wave born approximation ( dwba ) , we can write down the scattering amplitude from an initial channel @xmath48 where the target @xmath49 is in its ground state @xmath50 to a final channel @xmath51 where the residual nucleus is left in some excited state @xmath52 with excitation energy @xmath24 while the escaping proton has an energy @xmath53 . using the complex bi - orthogonal basis @xmath54 , we have : @xmath55 where @xmath56 is the interaction inducing the particle transfer from @xmath57 to @xmath49 , @xmath58 is an incoming wave of @xmath59 space at energy @xmath53 , and the sum over @xmath60 is restricted by the angular momenta and parities of states @xmath47 and @xmath61 .
the case where the final channel is the ground state of the residual nucleus corresponds to the above expression with @xmath62 .
the nucleon - transfer matrix element @xmath63 is proportional to the one - quasiparticle amplitude of the state @xmath47 . in analogy with the amplitude @xmath64 of eq.([f5 ] ) , we denote it by @xmath65 .
we also introduce the partial escape amplitudes of state @xmath47 to channel @xmath61 : @xmath66 the partial widths are : @xmath67 to obtain an expression for the cross section simple enough to lend itself to a discussion in terms of escape widths , let us assume furthermore that interference terms between different doorway states can be neglected .
the density of @xmath68 states is large , and for each interval centered around @xmath41 and containing @xmath69 states we define locally averaged quantities : @xmath70 then , one can rewrite the cross sections in the following form : @xmath71 the branching ratios can be calculated by the following formula : @xmath72 the above formalism is applied to study the semi - direct proton decay of the nuclei @xmath0sc , @xmath1cu and @xmath2nb with the aim of comparing the predictions with existing data from exclusive measurements .
although the qpm model soloviev et al.@xcite is not fully consistent since the residual interaction between quasiparticles is not derived from the quasiparticle mean field , it has the advantage that its two - body residual interaction is chosen of a multipole - multipole separable form and therefore , it enables one to work with configuration spaces of large dimensions without facing the cumbersome problem of diagonalizing large matrices .
this is very important for the results to be meaningful as it was already shown in a previous work on neutron emission@xcite .
the mean potentials entering the single - particle hamiltonians @xmath7 are of woods - saxon form .
their parameters are chosen according to refs.@xcite with some readjustments such that , once the coupling to phonons is included , the ground states are located at their respective experimental positions with respect to the proton separation threshold .
the qpm hamiltonian includes the monopole pairing which must be taken into account for nuclei with open shells@xcite .
this is the case when we calculate the core properties of @xmath1cu and @xmath2nb . the residual particle - hole interaction of eq . ( [ f2b ] )
is taken of a separable form in coordinate space with effective interaction strengths considered as adjustable parameters .
for the radial interaction form factor we have used @xmath73 where @xmath74 is the central part of the woods - saxon potential . for each excitation mode of the even - even core
corresponding to a given angular momentum , parity and isospin , the interaction strength is found by requiring that the lowest collective state calculated in rpa be at the experimental energy .
these interaction strengths are also used for the quasiparticle - phonon coupling .
we have modified the single - particle proton spectrum for @xmath2nb in comparison with ref.@xcite to get a better description of high - lying @xmath75 , @xmath76 subshells , but so far as energies of levels near the fermi surface are practically the same there are no changes in the properties of low - lying vibrational states in comparison with those of ref.@xcite .
for example , the calculated b(e@xmath77 ) values of the first low - lying collective states @xmath78 , @xmath79 in @xmath80ca are b(e3)=1.36@xmath81 , b(e5)=2.67@xmath82 that can be compared with the corresponding experimental values 1.24@xmath81 and 2.97@xmath82 . for all nuclei under consideration the calculated values of b(e@xmath77 ) are in general agreement with experiment . in the present model ,
these low - lying phonons of the core are the physical channels where the initial state in the excited odd - a nucleus can decay by semi - direct proton emission . in the actual calculation , a very large number of rpa phonons are included in the quasiparticle - phonon basis , but for @xmath80ca the first quadrupole state can not be treated as a one - phonon or particle - hole state and it is outside the present model . on the other hand , the experimental data @xcite show that the proton decay of @xmath0sc to the @xmath83 state in @xmath80ca is very weak . at high excitation energies ,
when the level density becomes large , it is often more convenient to calculate the strength function @xcite instead of solving the secular equations ( [ f10 ] ) .
one defines the strength function @xmath84 as the strength distribution @xmath85 folded with an averaging function @xmath86 : @xmath87 where @xmath88 is usually chosen as a breit - wigner function , @xmath89 .
it is shown in ref.@xcite that the calculation of the strength function @xmath84 can be done without the detailed knowledge of the amplitudes @xmath90 .
the same averaging procedure can be applied to calculate the partial escape widths and the branching ratios . in all calculations performed in this work
we have adopted the value @xmath91 = 0.1 mev .
using the methods presented above we have calculated the cross sections and branching ratios for the non - statistical proton decay of high - lying states with high angular momenta in @xmath0sc , @xmath1cu and @xmath2nb going to the ground and low - lying collective states of @xmath80ca , @xmath92ni and @xmath93zr , respectively .
an example of cross section calculations by the strength function method is shown in fig . 1 for the case of proton decay of the @xmath94 states in @xmath0sc .
since the calculated cross section is proportional to the square of the one quasiparticle amplitude of the wave function this figure enables one to see the @xmath95 strength distribution in @xmath0sc , too .
the @xmath95 strength is distributed over a broad energy interval due to the coupling with the collective vibrations .
results of our calculations for the @xmath95 proton particle strength and experimental data obtained from the @xmath80ca(@xmath96he , dp ) reaction at @xmath97=240 mev@xcite are presented in table 1 .
it is seen from this table that the calculations reproduce reasonably well the integral characteristics of the @xmath95 proton strength distribution in @xmath0sc .
the partial contributions of the non - statistical proton decay into the ground state and @xmath78 , @xmath79 states of @xmath80ca are presented in fig .
it is seen from fig .
2 that the partial cross sections are very energy dependent and as a result different channels can dominate at some excitation energies .
the solid curve in fig .
1 presents the sum of all 3 partial channels .
as one can see from fig . 2 at excitation energies below 8.4 mev the ground state channel dominates ,
but there are strong transitions to the @xmath79 state at 7.5 and 8.0 mev .
the @xmath78 and @xmath79 channels begin to contribute importantly starting from 8.4 mev and they become the main contributors after 9.5 mev . the sum of partial widths for different channels in some energy intervals are given in table 2 .
as one can see from table 2 for the energy interval 2.0 - 12.4 mev the proton decay to the @xmath79 state gives a contribution of about 54% and the @xmath78 channel contributes about 37% in the total sum .
there is a predominance of the @xmath79 channel in comparison with the @xmath78 one in the energy interval 7.0 - 12.4 mev .
such a behaviour can be understood from the structure of the decaying states and the angular momentum and energy carried away by the emitted proton .
it is seen from eq .
( [ w3 ] ) that the partial escape amplitude for fixed values of @xmath98 is proportional to the contribution of the quasiparticle - plus - phonon configuration made of a quasiparticle with the same @xmath98 and a phonon of the final state .
the cross section depends on the square of these amplitudes summed over all @xmath98 allowed by the angular momentum coupling rules . in the case of the proton decay of the @xmath94 states at excitation energies 712 mev into the @xmath78 channel the outgoing protons
carry @xmath60=(1,3/2 ) mainly whereas it is @xmath60=(1,3/2 ) and @xmath60=(1,1/2 ) for the @xmath99 channel .
thus , penetration factor arguments favour the latter channel .
it is well known that the strongest coupling between the single - particle states and phonons takes place for the collective phonons .
that is why the coupling with the low - lying vibrations is mainly responsible of the damping of the high - lying single - particle modes and giant resonances in nuclei .
this is the case for the @xmath78 and @xmath79 states in @xmath80ca . by this reason
the strengths of the @xmath78 and @xmath79 channels are redistributed in a broad energy interval .
the decrease of the spectroscopic strength with an increase of the excitation energy leads to a decreasing of cross sections of the channels discussed above .
it is worth to mention that a very similar behaviour for the @xmath78 and @xmath79 channels for the neutron decay of the high angular momentum states in @xmath3pb also takes place @xcite .
our conclusions about the role of different channels for the proton decay in @xmath0sc agree with observations made in ref.@xcite , but no separation of the direct part from the statistical one for the proton decay was done in that work . as a second example we consider the proton decay of the @xmath94 states in @xmath1cu .
results of our calculations for the @xmath94 strength distribution in @xmath1cu and experimental data@xcite are presented in table 3 .
the calculations reproduce correctly the strength near 7 mev and this part is mainly responsible for the proton decay . for the excitation energy interval ( 2 - 8.9 ) mev in @xmath1cu
the summed ground state width is equal to 10.3 ev and this is much less than in the @xmath0sc case .
an additional contribution from the 2@xmath100 channel has a summed width that is six times less than that of the ground state channel .
there are no transitions to other vibrational states .
this is easily understood if one looks at the structure of the states under discussion . for the @xmath101 channel
the outgoing protons carry @xmath60=(4,9/2 ) and there is a suppression of such transitions due to penetration factor effects .
the configurations including @xmath78 and @xmath79 states coupled with continuum are located at somewhat higher energies .
this is why no proton decay can proceed to the @xmath78 and @xmath79 states in @xmath1ni .
very recently , detailed experimental information about proton decay of isobaric analog states ( ias ) in @xmath2nb has been shown in ref .
the ias are strongly excited by means of the @xmath93zr@xmath102@xmath2nb reaction .
a sharp peak is seen at 12 mev excitation energy . around this
excitation energy three ias have been observed earlier@xcite at 11.93 mev , 12.07 mev and 12.15 mev .
it is shown in ref.@xcite that at 12 mev by means of @xmath102 reaction predominantly the @xmath103 state is excited .
the proton decays to the ground state and some low - lying states are observed .
the partial differential cross sections for each final state are given .
the data reveal proton decays predominantly to @xmath104 and @xmath105 excited states of @xmath93zr .
experimental data@xcite for energies and spectroscopic factors of ias and results of our calculations are given in table 4 .
besides , this table also contains calculated partial widths for the proton decay of three ias to the ground , 2@xmath106 , 3@xmath107 , 4@xmath106 and 5@xmath107 states
. as one can see from table 4 the calculated excitation energies for the @xmath108 ias are higher than the experimental ones but the spectroscopic factors extracted from experimental data within 20% accuracy are reproduced reasonably well .
let us discuss the proton partial widths for these ias .
it is seen from table 4 that for the first @xmath108 state the main decay channel is the @xmath109 channel . for this state
the @xmath60=(0,1/2 ) protons are most important for the @xmath109 channel and there is a rather weak transition for the @xmath83 channel due to the @xmath60=(2,3/2 ) protons .
the ground state width is proportional to the one quasiparticle strength ( the spectroscopic factor ) .
since the contribution of the one quasiparticle component in the norm of the wave function of this ias is much less than contributions of the quasiparticle - plus - phonon components the transition to the ground state is weaker than to the @xmath109 .
the particularities of the partial proton widths for the second @xmath108 state can be understood from the structure of this state , too . for the @xmath109 channel the @xmath60=(2,3/2),(2,5/2 ) and ( 4,7/2 )
protons can contribute besides the @xmath60=(0,1/2 ) protons . as a result
the decay width to the @xmath109 state increases strongly for the second @xmath108 ias in comparison with that of the first @xmath108 state .
an increase of the contribution of the configuration constructed from the first quadrupole phonon and @xmath110 single - particle state in comparison with their contribution to the structure of the first @xmath108 state results in an essential growth of the decay width to the @xmath83 channel .
the ground state width becomes larger mainly because of the increased spectroscopic factor .
the transitions to the @xmath104 state can take place due to the outgoing protons with @xmath60=(5,11/2 ) but they are supressed because of penetration factors and a small contribution of relevant components in the wave function structure .
in the case of the @xmath111 ias the @xmath112 channel dominates and this is due to the outgoing protons carrying @xmath60=(0,1/2),(2,3/2 ) and ( 2,5/2 ) .
the components containing the @xmath105 phonon have a small contribution in the wave function structure and as a result the width for the @xmath105 channel is small , too . in spite of unfavourable penetration factor
the @xmath83 width is 4 times larger than that of the @xmath105 channel .
this is due to the configuration consisting of the @xmath113 phonon coupled with @xmath103 which contributes about 15% in the norm of the @xmath111 ias .
the comparison of the calculated partial widths with the data of ref.@xcite shows that there is a qualitative agreement .
the main channels of the proton decay of ( 11/2)@xmath114 ias are reproduced in the calculated structure .
the dominance of the @xmath104 channel is well established and the calculated partial width of 3.3 kev ( table 4 ) is in agreement with the measured one ( 2.9 kev ) .
the @xmath105 channel is more pronounced than ground state channel but the calculated partial widths for both channels are much less than the measured ones .
the calculations indicate a large decay to the @xmath83 channel . such decay is discussed in ref.@xcite but quantitative estimations for the contribution of the @xmath83 channel in the cross section
have not been evaluated in that work .
the summed partial widths for the proton decay of the three ias to the low - lying vibrational states are also presented in table 4 . as one can see from the last row of table 4 the @xmath109 channel exhausts almost 60% of the total cross section , the @xmath104 channel contributes about 26% and the @xmath83 channel gives about 10% of the total strength .
the contribution of the ground state channel for the proton decay of the three ias in @xmath2nb is 4% only .
a microscopic approach based on the qpm has been applied to calculate the non - statistical proton decays of high angular momentum states excited in one - nucleon transfer reactions .
partial cross sections and branching ratios for proton emission from high - lying states in @xmath0sc , @xmath1ni and @xmath2nb have been evaluated .
the calculated branching ratios demonstrate the existence of strong direct proton decays to the low - lying vibrational states in @xmath0sc and @xmath2nb and enables one to understand particularities of the decay to different channels .
one can conclude from an analysis of calculated partial cross sections that , for high angular momentum states the non - statistical proton decay is more favourable into the higher angular momentum and lower excitation energy final states when the two following conditions are fulfilled : a strong particle - phonon coupling in that channel and a penetration factor which is not hindered by angular momentum or energy .
a similar conclusion was also reached for the neutron decay case@xcite . a general agreement with existing experimental data
is obtained and the predicted branching ratios for different channels can be used to analyze future experimental data .
we would like to thank g. crawley and s. fortier for fruitful discussions and correspondence .
ch.s . and v.v.v .
thank the hospitality of ipn - orsay where the main part of this work was done .
this work is partly supported by in2p3-jinr agreement and by the bulgarian science foundation ( contract no ph . | in the framework of the quasiparticle - phonon model , we study the non - statistical proton decay of excited states in odd nuclei towards low - lying collective states .
partial cross sections and branching ratios for the proton decay of the high angular momentum states in @xmath0sc , @xmath1cu and @xmath2nb are evaluated .
the calculated branching ratios predict strong direct proton decays to the low - lying vibrational states in @xmath0sc and @xmath2nb .
a general agreement with existing experimental data is found . |
cosmological @xmath2-body simulations of cold dark matter have revealed many empirical relationships concerning collisionless self - gravitating dark matter haloes , such as the density profile , the pseudo - phase - space density profile , the distributions of shape and spin parameters , the velocity - anisotropy - parameter profile and the distribution of velocity dispersion @xcite .
they used to be considered as being universal , i.e. different cold dark matter haloes obey the same empirical laws only with different scale parameters , which are independent of the halo mass , the initial density fluctuation spectrum and the values of the cosmological parameters .
the famous one of these universalities is the universal density profile of cold dark matter haloes .
these universal profiles are usually expressed as nfw @xcite or @xcite or even other specific forms , such as the @xcite profile .
we should notice that , however , more and more recent high - resolution @xmath2-body simulations claim that the dark halo density profiles are not strictly universal , which may depend on halo mass , merger history or substructures ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in fact , long before these simulation results of dark matter haloes were found , astronomical observations had already revealed that some self - bounded gravitating systems exhibit remarkable structural regularities .
for example , elliptical galaxies display a quasi - universal luminosity profile described by @xcite s @xmath3 law , and globular clusters are generally well fitted by the michie - king model @xcite .
although these empirical relationships are simple and practical , we just know what , but not why they behave in such ways , since they have no physical basis
. up to now , however , few of the physical origins concerning these empirical results have been investigated , except for the universal dark halo density profiles , whose origin was explored as soon as the universality was found @xcite .
thereafter , many other explanations for the origin of dark halo density profiles were proposed , and most of them , similar to the one by @xcite , are concentrated on various dynamical processes , such as collapsing , mergers of subhaloes or accretion of surrounding materials @xcite . however , a recent work by @xcite suspected this kind of dynamical origins . according to their simulations of hot dark matter haloes
, they claimed that mergers do not play a pivotal role in establishing the universalities , and thus their results contradict the models which explain the universalities as consequences of mergers .
is it possible that all these empirical relationships originate from statistical mechanics , or more concretely , do they reflect the properties of self - gravitating systems in statistical equilibrium states ?
if so , from theories of thermodynamics and statistical mechanics , we know that an equilibrium state of a thermodynamic system does not depend on any specific dynamical processes or mechanisms of how to attain it . if dark matter haloes really attain the statistical equilibrium states , then it would be very easy to understand the finding by @xcite , why the universalities are irrelevant to mergers .
the investigation of equilibrium statistical mechanics of isolated collisionless self - gravitating systems is a long - held issue , which can date back to the early 1960s .
@xcite proved that global entropy maxima do not exist for self - gravitating systems , which might be the first study that exhibits the great complexity of the statistical mechanics for self - gravitating systems .
later , @xcite in his seminal work proposed ` violent relaxation ' to account for the formation process of elliptical galaxies . just in the same work
, he derived a ` most probable ' distribution of a system in its ` equilibrium state ' that has infinite mass , energy and spatial extent , and hence is not realistic and not acceptable .
following @xcite s work , @xcite , @xcite and @xcite , with increasing complexity of phase - space partitions , reworked for their own coarse - grained distribution schemes .
these authors aimed to remedy the flaws in the previous works , the most significant of which is the mass segregation problem , i.e. in a multi - species composition , the heaviest objects tend to concentrate more centrally whereas the lightest ones tend to go towards the border .
however , these authors did not overcome the problem of infinite mass that arises from @xcite s statistical approach .
we refer the reader to @xcite for the comprehensive review about the history of these works .
since the concept of entropy is the basis of statistical mechanics , ever since @xcite , who used the usual boltzmann - gibbs entropy , various alternative entropy forms have been attempted to account for the stability and equilibrium configurations of self - gravitating systems , such as the generalized @xmath4-function @xcite , the entropy of ideal gas @xcite or tsallis entropy @xcite .
there are also other authors using entropy arguments to draw conclusions about the final states of violent relaxation ( e.g. * ? ? ?
* ; * ? ? ?
all these entropies are maximized , subject to the usual mass- and energy - conservation constraints @xcite , or other additional conditions , such as hydrostatic equilibrium @xcite .
usually , the equilibrium state of a thermodynamic system is defined by employing the principle of maximum entropy , which states as follows : the realistic thermodynamic relations of the stable equilibrium state for any thermodynamic system may be derived by seeking the probability density function that maximizes the boltzmann - gibbs entropy ( or other plausible entropy forms ) , subject to all relevant constraints .
however , the principle of maximum entropy for self - gravitating systems has been continually questioned for the past several decades .
for example , there is a famous argument , originated by j. binney , against the feasibility of boltzmann - gibbs entropy , in that such an entropy form is unbounded above ( @xcite ; see also @xcite ) , which resembles @xcite s earlier result that there exists no global maximum entropy for self - gravitating systems .
confirming an earlier study by @xcite , @xcite showed that the general @xmath4-functions during the process of violent relaxation , may not necessarily increase at all times and may decrease if the system has specific , far from equilibrium , conditions .
they further proposed to distinguish between two phases of violent relaxation , i.e. in one phase , the @xmath4-function may decrease with time , whereas in the other it is always a non - decreasing function of time .
this result suggests that whether @xmath4-function is a monotonically increasing function of time during the violent relaxation process is still an open question @xcite , or equivalently , the equilibrium state of self - gravitating systems should not be simply considered as the maximum entropy state @xcite .
worse still , @xcite revealed that there exists an inconsistency in theories of violent relaxation with @xcite and @xcite s distributions .
the inconsistency arises from the non - transitivity of these theories , i.e. a system that undergoes a violent relaxation , then followed by another , with an addition of energy for the latter , is not equivalent to the one that undergoes the whole process directly from the initial to the final state .
the non - transitivity is destructive , since the ` equilibrium state ' is dependent on the evolutionary path of the system , which implies that a complete methodology of statistical mechanics is probably impossible for self - gravitating systems , and should be replaced by some dynamical approaches .
therefore , after unsuccessful attempts for nearly 50 yr , the concept of entropy as well as the methodology of statistical mechanics might have been widely regarded as inviable for self - gravitating systems . as a result , we are in such an embarrassing situation that we have few applicable analytic theories concerning self - gravitating systems , such as elliptical galaxies and dark matter haloes , except for the poisson - vlasov equations ( or the jeans equation ) and the fitting formulae from either @xmath2-body simulations or observations . at the same time , these problems and difficulties agonize statistical physicists as well .
they raised such questions for the general long - range interaction systems , as non - additivity or non - extension , ensemble inequivalence and negative specific heat in microcanonical ensemble and ergodicity breaking @xcite .
moreover , since it is not known how to define the equilibrium state of a self - gravitating system , such a state that an equilibrated system attains is vaguely called the quasi - stationary state .
it seems that , in order to establish a successful statistical mechanics of self - gravitating systems , some new ideas or even conceptual revolutions may be desired to get out of this long - term impasse . in this work
, we explore whether the puzzle concerning the equilibrium statistical mechanics of self - gravitating systems can be solved by introducing some new thoughts , or by revising some fundamental laws or principles of physics .
our paper is arranged as follows . in section
[ sec : basic ] , we firstly propose a mean - field model to separate gravitational forces into long- and short - range parts , and subsequently identify three distinct relaxation mechanisms , i.e. violent relaxation , two - body relaxation and gravitational landau damping .
then by calculating the first - order variation of the entropy of @xcite , we derive an entropy stationary equation . in section [ sec : result ] , combined with the jeans equation , we solve the two equations in two cases .
first , we show that the velocity anisotropy parameter is of great significance to make the density profiles of dark matter haloes have finite mass , energy and extent . then , ignoring the gravitational degeneracy in the inner region of the dark matter halo , we find that the predicted anisotropy parameter and the pseudo - phase - space density agree rather well with the simulations of @xcite in the outer non - degenerate region , and are also in rough agreements with the simulations in the middle weak - degenerate region of the dark halo . an upturn deviation from the power law of the pseudo - phase - space density at the halo outskirts is also predicted , which is consistent with the finding by @xcite . in section [ sec : gen2nd ] , we generalize the standard second law of thermodynamics to long - range self - gravitating systems , which states in the complementary way as follows : for an isolated self - gravitating system , the entropy can _ automatically decrease _ under long - range violent relaxation , until violent relaxation terminates and the system attains the equilibrium state with the global _ minimum _ entropy for the whole system ; yet , in this very equilibrium state , the system is simultaneously in the local _ maximum _ entropy state for every and any part of the system under short - range two - body relaxation and landau damping . in section [ sec
: cons ] , we conclude that , with this revised , complementary second law of thermodynamics , the concept of entropy is still valid , and thus in principle , the equilibrium statistical mechanics can also be established for such long - range self - gravitating systems .
it is usually believed that , undergoing the violent relaxation procedure @xcite , a self - gravitating system can attain a so - called quasi - stationary state @xcite , i.e. the virial equilibrium state . throughout this paper
, we tentatively call this quasi - stationary state as equilibrium state . the appropriate gravitational model should be the mean - field model , which is the first step towards the correct statistical mechanics of self - gravitating systems . the following is a brief introduction to the idea of mean - field model , which was suggested by @xcite . for the detailed discussion of the mean - field model for self - gravitating systems ,
we refer the reader to their original article .
there are also many discussions about the mean - field model that can be found in @xcite . usually , the standard statistical mechanics deals with short - range interaction systems .
the typical feature of these systems is that the interaction between two particles is strong and repulsive when they are very close to each other , but on large distances no force is exerted between two particles . as a result , particles most of time move at nearly constant velocity , and then are subject to violent and instantaneous accelerations when they collide with one another . in this situation ,
the typical relaxation mechanism arises from two - body collisions .
one of the difficulties of developing the statistical mechanics of self - gravitating systems lies in that gravitational interaction is long - range attractive force , which means that a particle is coupled to almost all other particles at all spatial scales of the system .
so it seems that all different range of scales are equally important to the dynamical properties of self - gravitating systems .
fortunately , we can circumvent this obstacle by constructing a gravitational model , in which the long - range part of gravitational force between the particles is collectively manifested as a mean field for a finite self - gravitating system , and the short - range part arising from the nearby particles is treated as fluctuations in this mean gravitational field .
note that the equilibrated self - gravitating systems usually possess some geometrical symmetry , by virtue of which the long - range coupling between the particles can be broken when the system is in its equilibrium states .
we illustrate this separation of gravitational force with a simplified one - dimensional model as follows .
see fig .
[ fig : fig1](a ) , for the one - dimensional system in equilibrium , the particles are homogeneously distributed in the mean field , indicated by the dashed line , with a mean separation as @xmath5 . in this model ,
if a particle is initially located at point @xmath6 , it will not feel any gravitation . when the particle moves towards the right side , however , it will be more and more affected by the gravitational force of particle @xmath7 .
when the particle crosses over @xmath7 on the right side , continuously moving farther and farther away , the gravitational force exerted on this particle becomes less and less with the increasing distance from @xmath7 , and vanishes again at point @xmath8 .
so from this model , we can see that after removing the mean - field force , the effective short - range part of gravitational force of a particle is just dictated by its nearby particles .
we separate the gravitational potential and the density as @xmath9 , and @xmath10 , in which @xmath11 , @xmath12 are the long - range mean - field quantities , and @xmath13 , @xmath14 are the short - range residuals , respectively . with poisson equation @xmath15 ,
these long- and short - range dynamical variables satisfy their respective poisson equations @xmath16 thus , we can see that , in this mean - field model , the mean , smoothed - out gravitational potential is treated as a background field , corresponding to large - scale perturbations ; whereas the short - range residual effects are regarded as fluctuations in this background , corresponding to small - scale perturbations . in the mean - field model , a particle couples not directly with particles at large separations , but only with this background potential , with the residual short - range effects determined by a few particles nearby . in later sections , we will see the great significance of this separation of gravitational force , as well as different physical implications by distinguishing between these two types of perturbations .
it is the instability of the mean gravitational potential , i.e. @xmath17 , that induces a strong , long - range relaxation mechanism , called violent relaxation @xcite .
when violent relaxation ceases , i.e. @xmath18 , it leaves behind for the system the virial equilibrium state , @xmath19 , or equivalently , a hydrostatic equilibrium state , which is described by the jeans equation ( see * ? ? ?
* ) @xmath20 where the ` pressure ' @xmath21 is defined as @xmath22 , and @xmath23 is the radial velocity dispersion .
@xmath0 is the velocity anisotropy parameter , usually defined as @xmath24 , where @xmath25 is the one - dimensional tangential velocity dispersion , with @xmath26 .
the short - range residual effects of the gravitational forces between these particles are manifested in the following two forms of relaxation mechanisms .
one is two - body relaxation , which arises from encounters of two bodies , and is only efficient locally , or at small scales of the system . at first
, the system stays at its maximum radius with vanishing kinetic ( or thermal ) energy , and then is driven by violent relaxation to approach the virial equilibrium state .
it is well known that , at this stage , two - body relaxation operates in a far longer time - scale than violent relaxation , so that it should actually be neglected when violent relaxation is functioning .
the other short - range relaxation mechanism operating at small scales in a self - gravitating system is gravitational landau damping , which was first realized by @xcite and further studied by @xcite .
a brief introduction to this short - range relaxation is the following .
a density perturbation with wavelength shorter than the jeans length , @xmath27 , will induce the jeans waves ; otherwise the perturbation will induce the jeans instability .
such jeans waves in self - gravitating systems , heavily damped by interactions with particles , are the counterparts of langmuir waves in plasma . from the dispersion relation of the self - gravitating systems ( equation 5.66 and fig .
5.2 of @xcite ) , we can see that landau damping is also very efficient locally .
hence , the two short - range relaxation mechanisms are responsible for the local equilibria of the system at small scales .
these relaxation mechanisms are picturized in fig .
[ fig : fig1](b ) .
we can see that the three relaxation mechanisms , violent relaxation , two - body relaxation and landau damping operate at different evolutionary stages and different spatial scales of the system : ( 1 ) two - body relaxation , which arises from interactions between particles themselves , is a short - range relaxation mechanism ; ( 2 ) landau damping , which is another short - range relaxation , arises from interactions between particles and the jeans waves ; ( 3 ) whereas if perturbations take place in large - scale regions in the early evolutionary stage , such large - scale perturbations will induce long - range violent relaxation , i.e. the large - range violent changing of the gravitational potential .
we will see later the significance of the discrimination between these three types of relaxation mechanisms .
after violent relaxation terminates , the system stays in the hydrostatic equilibrium state , described by the jeans equation , equation ( [ eq : je ] ) . in this section
, we have interests to investigate whether there are some extra physics resulting from violent relaxation besides the hydrostatic equilibrium .
as mentioned previously , there have been great disputes about the validity of the concept of entropy for self - gravitating systems for nearly five decades .
ignoring the controversy , we tentatively employ the entropy principle to explore the possible thermodynamic relations of self - gravitating dark matter haloes .
we choose the specific entropy in the phenomenological form as the following @xmath28 which was first used by @xcite for the isotropic self - bounded gas sphere , but here we assume it can also apply to the velocity - anisotropic case .
this expression is the standard entropy form of an ideal gas in terms of @xmath21 and @xmath29 , without the normalizing factor and an additive constant , and we borrow it for self - gravitating systems without any a priori reasons . with the specific entropy given ,
the total entropy of the system is @xmath30 as usual , the constraints are the total mass and energy conservation .
the mass function @xmath31 of a spherical gravitating system is @xmath32 and the total mass of the system is @xmath33 for the convenience of variational calculus , we set @xmath34 .
so the mass - conservation constraint is converted to satisfying the following boundary conditions : @xmath35 and we replace @xmath36 in the entropy functional with @xmath37 , the first derivative of @xmath38 , as @xmath39 the total binding energy of the system with spherical symmetry is @xmath40 so by using the virial theorem , the total energy of the system is @xmath41 the constrained entropy is @xmath42 where @xmath43 is the lagrangian multiplier .
substitute equations ( [ eq : tots ] ) , ( [ eq : mf ] ) , ( [ eq : rhod ] ) , and ( [ eq : et ] ) into equation ( [ eq : cs1 ] ) , we have @xmath44 { { \rm d}}r.\end{aligned}\ ] ] we treat @xmath21 as a known function of @xmath45 , and only calculate the first - order variation of @xmath46 with respect to @xmath47 and @xmath48 . we obtain the following euler - lagrangian equation from @xmath49 as @xmath50 since this equation comes from the first - order variational calculus , in what follows , we call equation ( [ eq : seq ] ) as the entropy stationary equation , or simply , the entropy equation .
additionally , from equation ( [ eq : specs ] ) , it is interesting to notice that @xmath51 that is , the so - called pseudo - phase - space density @xmath1 seems to be related with the specific entropy @xmath52 , which implies that the pseudo - phase - space density may really have a physical origin .
this relation has been recognized by ( * ? ? ?
* see equation 7.121 ) .
@xcite also suggested such a correlation between them that the phase - space density profile likely reflects the distribution of entropy , which dark matter acquires as it is accreted on to a growing halo . with a simple dimensional analysis
, we can write the lagrangian multiplier @xmath43 in equation ( [ eq : seq ] ) as @xmath53 with @xmath54 being a positive dimensionless number , so that @xmath43 is always negative . in the following
, we can see that equations ( [ eq : je ] ) and ( [ eq : seq ] ) have no solutions if @xmath43 is positive .
we can always set @xmath55 by choosing appropriate values for the characteristic density @xmath56 and scale @xmath57 .
in this work , we use the radial pseudo - phase - space density @xmath1 instead of the more general @xmath58 to formulate the specific entropy , which by no means indicates that the former is physically superior to the latter . in fact
, we have tried many combinations of @xmath59 , but all failed to yield reasonable results .
for example , take the form of @xmath59 as @xmath60 , and perform the same variational calculus as for equation ( [ eq : seq ] ) , we get the equation @xmath61 or with @xmath62 , we derive the following result : @xmath63 compared with the entropy stationary equation ( [ eq : seq ] ) , it can be seen that there appears an additional term , @xmath64 , or @xmath65 in the above two equations . from fig . 9 of navarro et al .
( 2008 ) , we can see that @xmath0 changes sharply at the outskirts of the dark matter halo , so the absolute value of @xmath66 at the outer boundary is huge .
such an additional term containing @xmath0 will greatly affect the behaviour of the solution of equations ( [ eq : seq2 ] ) or ( [ eq : seq3 ] ) , so that we can not even get a convergent solution .
moreover , there is also another problem that we can not perform the variational calculus with respect to the pressure @xmath21 , since such a resultant equation is meaningless and unacceptable , so in the current work we have to treat @xmath21 as a known function . as a consequence , such a treatment is inconsistent , and hence at this moment we can obtain only one equation by using the entropy principle . because of this inconsistency , our results are incomplete .
these problems indicate the great complexity of the statistical physics of self - gravitating systems , and imply that , either the entropy form may be even more complicated than the current forms we have already considered , or we may still lack some crucial physics , such as some extra macroscopic constraints besides the mass and energy conservation .
we believe these problems should be the targets of future investigations .
up to now , for the three independent variables @xmath21 , @xmath29 and @xmath0 , we just have two equations , the jeans equation ( [ eq : je ] ) , and the entropy equation ( [ eq : seq ] ) . in order to obtain definite solutions of them
, we have to provide an extra equation to close the equation system .
this can be done by assigning one of the three variables with a known function . in the following ,
we explore the solutions in two cases , respectively : ( 1 ) given the anisotropy parameter @xmath0 , and ( 2 ) given the density profile @xmath29 .
with the anisotropy parameter @xmath0 given , the density profiles can be solved , and these solutions are shown in fig . [
fig : fig2 ] . for simplicity , we specify @xmath0 as a step function of @xmath45 as @xmath67 in which @xmath68 is arbitrarily specified as , say @xmath69 , and the values of the constant @xmath70
are indicated in fig .
[ fig : fig2 ] .
first , we concentrate on the case of @xmath71 . here , with the lagrangian multiplier @xmath43 chosen to be negative , the two equations have neat analytic solutions as @xmath72 and @xmath73 , which are exactly the isothermal solutions .
it is well known that in such a solution , the system has infinite mass , energy and extent , which is similar to @xcite s result , and so is not acceptable .
such an unconfined density configuration of self - gravitating systems once misguided @xcite and @xcite to believe that violent relaxation is not a complete relaxation mechanism , and so introduce the quest for a theory of incomplete relaxation ( see also * ? ? ?
however , we can see that cases with @xmath74 are different from the isothermal solution .
these solutions gradually deviate downwards from the isothermal solution and descend more and more rapidly with increasing radius .
the larger the value of @xmath0 , the more decline of the solution , and the density functions drop down sharply at some outer radii .
it is obvious to see that @xmath0 is of great significance to make the solutions have finite mass , energy and spatial extent , and therefore , with @xmath0 involved , it is not necessary to introduce a phase - space cut - off , or incomplete relaxation mechanisms @xcite . ) and ( [ eq : seq ] ) , with @xmath0 being a given function of @xmath45 in equation ( [ eq : beta ] ) .
the fiducial solution of @xmath71 is indicated as ` isothermal ' in the figure ; others are shown as @xmath75 . ]
now let us discuss the other case , i.e. solving equations ( [ eq : je ] ) and ( [ eq : seq ] ) by specifying a dark halo density profile @xmath36 . here , we use the @xcite profile ( also see * ? ? ?
* ) , @xmath76,\ ] ] with the shape parameter @xmath77 , the same as the one taken by @xcite for their aq - a halo simulation series .
we solve the two equations from the outer boundary towards the inner of the dark matter halo .
the two solutions are shown as @xmath78 and the radial pseudo - phase - space density @xmath1 in fig .
[ fig : fig3 ] . compared with the corresponding simulation results in (
* see figs 9 and 12 ) , we can see that our solutions in the inner part of the dark matter halo are completely unacceptable .
we ascribe this disagreement between our theoretical predictions and the simulation results to the phenomenon called gravitational degeneracy @xcite , which is not included in this paper .
we briefly introduce this effect in the following .
@xcite derived a distribution of self - gravitating stellar systems that is analogous to the fermi - dirac distribution . in his work ,
the gravitating particles are grouped into , and represented by the coarse - grained , constant - density phase - space elements , which have fixed and finite phase volume , and can not overlap to each other .
it is this fermi - dirac - type distribution that leads to the existence of the gravitational degeneracy in the inner region of self - gravitating systems , where is much denser than the other parts .
such a gravitational degeneracy provides an exclusive effect against the gravitational collapse to avoid the gravothermal catastrophe @xcite .
it should not be confused that the gravitational degeneracy results not from pauli exclusion principle of quantum mechanics , but from the conservation of the overall phase - space volume of these phase elements . in strong degenerate region ,
i.e. the inner part of the dark halo , the close encounters should not be ignored so that the specific entropy , equation ( [ eq : specs ] ) , which is just suitable for ideal gas composed of point - like particles , is not proper .
it is valid only in the non - degenerate limit , i.e. the outer low - density region of the dark halo . as one of the solutions of equations ( [ eq :
je ] ) and ( [ eq : seq ] ) , with the density profile given as the einasto profile in equation ( [ eq : einasto ] ) .
we numerically integrate the two equations towards the inner of the dark matter halo , from the outer boundary at @xmath79 , where the logarithmic slope of the einasto profile is @xmath80 .
the initial values of @xmath0 are shown in the figure .
the arrow indicates the position of the local peak , at about @xmath81 .
( b ) the other solution of equations ( [ eq : je ] ) and ( [ eq : seq ] ) , shown as the radial pseudo - phase - space density profile @xmath1 as functions of radius @xmath45 .
the vertical scale is unnormalized . ]
if the degenerate effect is included , the proper specific entropy may be better generalized to , say , the @xcite s form @xmath82 where @xmath83 is the pseudo - phase - space density , and @xmath84 is the maximum phase - space density , which embodies an exclusive effect that prevents different phase elements from overlapping to each other in the final state . in the current work , however , we restrict ourselves not to include this complexity of gravitational degeneracy . according to the behaviour of the solutions , we draw a vertical dotted line at @xmath85 in fig . [
fig : fig3 ] , and an arrow pointing to the local peak at @xmath86 in the upper panel , to roughly separate the halo into the innermost strong - degenerate ( with the radius @xmath87 ) , the middle weak - degenerate ( @xmath88 ) and the outmost non - degenerate ( @xmath89 ) region . as mentioned above , due to the ignoring of gravitational degeneracy , our theoretical predictions in the inner strong - degenerate region of the halo are not reliable at all .
so in the following we just concentrate on the solutions in the outer regions .
the velocity - anisotropy parameter for dark matter haloes that is greater than for baryons has been observationally conformed @xcite , whereas the latest high - resolution simulations revealed much more details that are absent in the observations . @xcite and @xcite found the non - monotonic radial dependence of @xmath0 , i.e. haloes are nearly isotropic near the center , radially biased to the maximum at some radius and approximately isotropic again in the outskirts , which is against the earlier result of @xcite , who found that the anisotropy parameter just increases monotonically outwards from the inner part of the dark matter halo . from fig .
[ fig : fig3 ] , we see that the predicted @xmath0 agrees quantitatively well with the corresponding simulation results of @xcite in the non - degenerate region .
in particular , it is worth emphasizing that our theoretical predictions completely reproduce the non - monotonic radial dependence of @xmath0 outside the strong - degenerate kernel . besides , our predictions are also in qualitative agreement with the simulation results in the middle weak - degenerate region of the dark halo .
similar to the case of @xmath0 , the predicted pseudo - phase - space densities @xmath90 in the inner strong - degenerate kernel also greatly deviate from the simulated results , for the same reason as that for @xmath0 , i.e. ignoring of the gravitational degeneracy . in the outer regions with @xmath91 , however , the predictions , similar to the simulation results , can also be well fitted by @xmath92 the power law predicted by @xcite s similarity solution for infall onto a point mass in an otherwise unperturbed einstein - de sitter universe except for an upward deviation in the halo outskirts .
we are delighted to see that our result is in good agreement with the finding by @xcite , that such a curve - up deviation also exists in their simulations .
this significant upward deviation from the simple power law of @xmath92 indicates that the great complexity exists in equilibrated dark matter haloes @xcite .
thus , we have seen that , compared with the simulations , our theoretical results provide good agreements in the outer non - degenerate region of the dark halo , and also acceptable agreements in the weak - degenerate region , respectively .
we conclude that these agreements between our theoretical predictions and simulation results sufficiently indicate that both the concept of entropy and the entropy principle should be still valid for long - range self - gravitating systems . ;
( b ) small - scale density perturbation / fluctuation , i.e. @xmath93 , where @xmath6 and @xmath8 are defined in equation ( [ eq : ab ] ) . ]
we would like to inspect whether the solution of the first - order variational calculus , i.e. equation ( [ eq : seq ] ) , is a maximum , minimum or any other types of solutions .
for this sake , we calculate the second - order variation . by straightforward substitution ,
the second - order variation of the constrained entropy of equation ( [ eq : cs2 ] ) is @xmath94 { { \rm d}}r \nonumber \\ & = & \frac{1}{2 } \int^{\infty}_0 [ a(\delta m)^2 + b(4\pi r^2 \delta \rho)^2 ] { { \rm d}}r,\end{aligned}\ ] ] where @xmath95 for the convenience of the following analysis , we call the first term in the above integral , @xmath96 , as _ mass perturbation _ , and the second term , @xmath97 , as _ density perturbation_. according to the result of the first - order variational calculus , we know @xmath98 , so we can immediately see that @xmath99 , and @xmath100 .
this result indicates that the entropy stationary solution , equation ( [ eq : seq ] ) , is neither a maximum nor a minimum , but a saddle - point solution of @xmath49 .
so at first sight , it seems that the solution might be unstable , and the entropy form of equation ( [ eq : specs ] ) may not be proper for self - gravitating systems .
in fact , this apparent catastrophe contains a profound new physics that we may have never realized before . in what follows
, we will analyze it in detail .
first , we assume that @xmath101 , or equivalently , @xmath102 here , the mass perturbation is predominant , whereas the density perturbation , i.e. the variation of the first derivative of mass function , is negligible . as a result
, such a mass perturbation must be large - scale , which is shown in fig .
[ fig : fig4](a ) .
one may ask what physical processes can cause such kind of mass perturbation .
a rough analysis shows that the dynamical processes , such as gravitational collapse or the system s radial large - range oscillations with large amplitude , can all lead to the mass perturbations , which must induce the violent changing of the long - range , background gravitational potential with respect to time , i.e. @xmath17 .
this is exactly the feature of violent relaxation .
so from equation ( [ eq:2ndvar ] ) , with @xmath99 and without small - scale density perturbation , we see that the entropy of the isolated self - gravitating systems , driven by violent relaxation , will always _ automatically decrease _ to the minimum until violent relaxation terminates .
furthermore , we notice that the mass perturbation , @xmath103 , does not contain any localized information , say , some function of the density @xmath36 , hence we can further identify this equilibrium entropy as the global minimum entropy .
our result is consistent with binney s argument , which was designed to argue against the validity of boltzmann - gibbs entropy as well as the maximum entropy principle for self - gravitating systems .
this argument briefly states as follows .
the total mass and total energy of the system are assumed to be @xmath104 and @xmath105 ( negative ) , respectively .
a large fraction of the mass and the energy , say @xmath106 and @xmath107 , are concentrated in a bound static core , whereas the remaining mass @xmath54 and thermal energy @xmath108 are deposited in a diffuse outer envelope . as @xmath109 the envelope becomes larger and larger in radius , but its density becomes more and more tenuous , such that the associated entropy will diverge to infinity @xcite . so the fact that entropy is unbounded above is consistent with our finding .
besides binney s argument , @xcite also proved that global maximum entropy states do not exist for self - gravitating systems ( see also appendix iii of @xcite ) , which is also a support to our conclusion .
now we turn to the analysis of density perturbation .
here , we have @xmath93 , or equivalently , @xmath110 in the context of density perturbation , the mass perturbation is negligible , which implies @xmath111 , i.e. there is no violent relaxation .
so we can see from fig .
[ fig : fig4](b ) that the density perturbation is just small - scale perturbation .
such small - scale density perturbations will induce only short - range relaxations , i.e. @xmath112 , such as two - body relaxation and landau damping . with @xmath100 and without mass perturbation
, we can see from equation ( [ eq:2ndvar ] ) that the entropy is the maximum for the system under short - range relaxations .
moreover , because of the small - scale nature of the density perturbation , we can further identify this equilibrium entropy as the local maximum entropy , not for the whole entity , but just localized for every and any small part , or roughly , coarse - grain , of the system .
we further illustrate the differences of the long- and short - range relaxation mechanisms in fig .
[ fig : fig5 ] . when the system has experienced a large - scale mass perturbation and departures greatly from the equilibrium state
, long - range violent relaxation will be excited and drive the system back to the equilibrium state , i.e. the minimum entropy state . nevertheless ,
if violent relaxation nearly or completely terminates , then short - range two - body relaxation / landau damping operate locally to maintain the local maximum entropy state for every / any small part of the system .
in fact , in the usual thermodynamics with no long - range relaxation mechanisms , if the system is extremely large and unconfined , such as the atmosphere of the earth , it is impossible to acquire a global , but just local thermodynamic equilibria , for the system .
so we can see that things are indeed complicated for the self - gravitating system , but our results do not conflict with the standard second law of thermodynamics at all , since the latter is just valid for the systems dominated by short - range relaxation mechanisms .
we here just generalize the second law to accommodate the long - range , self - gravitating systems . to summarize , we re - express the second law of thermodynamics for self - gravitating systems in the seemingly paradoxical , but actually complementary way as follows : for an isolated self - gravitating system , if long - range relaxation ( i.e. violent relaxation ) is predominant , the entropy of the system never increases ; whereas if short - range relaxations ( i.e. two - body relaxation together with landau damping ) are predominant , the entropy never decreases .
an additional explanation for this revision of the second law of thermodynamics states that the equilibrium states of self - gravitating systems are the global minimum state for the whole system under long - range relaxation , but the local maximum entropy state for every and any small part of the system under short - range relaxations .
it is only for the long - range self - gravitating systems that there exists such a minimum - maximum entropy duality , and the corresponding complementarity for the second law of thermodynamics . for the usual thermodynamic systems
, there is no distinction between long - range and short - range relaxations , so there exists no such a complementarity for the standard second law of thermodynamics . as addressed in the introduction , some earlier authors ( e.g. * ? ? ?
* ; * ? ? ?
* ) have already realized that the @xmath4-function , or entropy of self - gravitating systems , may not necessarily increase during the violent relaxation process .
@xcite even distinguished between two phases of violent relaxation , in which the @xmath4-function may decrease with time in the first phase , i.e. the ` vigorous phase ' they called , whereas in the second phase , i.e. the ` calm phase ' they called , it is always a non - decreasing function of time .
if we tentatively regard landau damping as the second phase of violent relaxation as they referred to , one can notice that their finding is very close to our result , which is a great support to the complementarity we found in this work .
however , compared with our distinction between the three relaxation mechanisms , their two - phase classification seems slightly vague and inaccurate , and they did not relate these two phases to the long- and short - range relaxations , either .
most of all , it seems that they did not realize the new physics , i.e. the complementary second law of thermodynamics , underlying such a decreasing / incresing property of the @xmath4-function .
we propose a thought experiment in fig .
[ fig : fig6 ] , which may help to understand this generalized , or complementary , second law of thermodynamics .
the processes are the following .
( i ) an ideal gas is confined within an adiabatic box , which resides in the vacuum .
the short - range relaxations , namely , collisions of the particles between themselves and with the walls of the box , maintain the ideal gas in its thermal equilibrium , i.e. the maximum entropy state .
( ii ) suddenly remove the walls , and the gas begins to expand outwards .
( iii ) this is just the familiar free - expansion process of an ideal gas , in which the gas will expand all the way without ending , and we know that the entropy will increase to infinity along with the expansion .
now we time reverse the above processes from ( i ) to ( iii ) , and then the time - reversal processes proceed in this way : ( iv ) the gas moves inwards to concentrate to a small spatial region , pulled by some fictitious ` long - range attractive force ' , and of course the entropy of the gas decreases with this process .
( v ) the ideal gas is suddenly enclosed by adiabatic walls , and then the ` long - range relaxation ' stops , leaving behind a mechanical equilibrium of the ideal gas .
the entropy of the system also stops decreasing , so we may say the system attains its minimum entropy state . ( vi ) the equilibrated ideal gas implemented by the ` long - range relaxation ' is then maintained by the same short - range relaxations as in stage ( i ) , so that the entropy is the maximum under the short - range relaxations .
we can see that , except for the sudden establishing of the potential well , i.e. the adiabatic walls , this time - reversal procedure greatly resembles the gravitational collapsing process under the action of gravitational force .
this thought experiment helps to understand the seemingly strange behavior of the minimum
maximum entropy states of self - gravitating systems .
the statistical mechanics of self - gravitating systems has long been believed as an inviable discipline .
the controversy mostly focuses on whether the concept of entropy is still valid for long - range interaction systems , such as self - gravitating systems . in this work
, we perform a preliminary study to this age - old problem .
we summarize our findings in the following .
when self - gravitating systems attain their equilibrium states , the systems generally possess some geometrical symmetry , such as a spherical shape . under this condition , it is desirable to approximate the self - gravitating systems by the mean - field model .
that is , the long - range part of gravitational forces between the particles is collectively manifested as a smooth background force field , with the short - range residual effects determined by nearby particles .
such a separation of long- and short - range gravitational force breaks the long - range coupling of a particle with the others at all different range of scales , and definitely renders the problem more manageable .
entropy is a state function that is introduced to describe the degree of order or chaos of a thermodynamic system .
this concept , however , is of great significance in statistical mechanics , not just because it is a thermodynamic state function , but also because it constitutes a variational principle called the entropy principle ( with the obsolete name ` principle of maximum entropy ' ) , from which the equations or relations of thermodynamic quantities can be derived .
it plays the same role for statistical mechanics as the one played by the principle of least action for fundamental dynamics . without an appropriate definition of entropy of the thermodynamic system under consideration , it is impossible to establish the statistical mechanics for this system , as has been exactly the case of self - gravitating systems for the past several decades . in this work ,
we tentatively employ the phenomenological entropy form of ideal gas , first used by @xcite , to explore the possible thermodynamic relations of the equilibrium states of self - gravitating systems . by calculating the first - order variation with respect to the density function @xmath29 , subject to the constraints of mass and energy conservation , we derive an entropy stationary equation .
however , we acknowledge that our treatment is somewhat deficient in completeness or consistency , in that we have to treat the pressure @xmath21 as a known function , and can not calculate the variation with respect to it . as a result , we can derive only one equation , i.e. the entropy stationary equation .
moreover , we have also tried some other entropy forms other than the @xcite form , but all failed to yield any reasonable results .
these problems imply that either the entropy form may be even more complicated than the current forms we have already considered , or we may still lack some crucial physics , such as some extra macroscopic constraints besides the mass and energy conservation . however , despite this incompleteness / inconsistency , when incorporated with the jeans equation , we do find some interesting results with this entropy equation . in the case of given anisotropy parameter @xmath0 , we solve the entropy stationary equation and the jeans equation to obtain @xmath21 and @xmath29 . if @xmath113 , the two equations possess the isothermal solutions .
solutions with @xmath74 deviate downwards from the isothermal solution , descend more and more rapidly with increasing radius , and drop down sharply at some outer radii .
we conclude that @xmath0 is of great significance to make the solutions have finite mass , energy and spatial extent , and hence it is not necessary to introduce a phase - space cut - off , or incomplete relaxation mechanisms . according to the behaviour of the solutions , we separate the halo into the inner strong - degenerate ( @xmath114 ) , the middle weak - degenerate ( @xmath115 ) and the outer non - degenerate ( @xmath116 ) region . in the case of given density profile @xmath29 ,
say the einasto profile , we derive the solutions of @xmath0 and @xmath1 , which are in good agreements with the simulation results of @xcite in the outer non - degenerate region of the dark halo , and also in rough agreements with the simulations in the middle weak - degenerate region .
in particular , it is worth emphasizing that our theory predicts an upward deviation of @xmath1 from the power law , @xmath117 , at the halo outskirts , which is consistent with the finding by @xcite , that such a curve - up deviation also exists in their simulations .
however , we can not explain the simulation results in the inner strong - degenerate region , since we have not considered the effect of gravitational degeneracy in the current theory . by calculating the second - order variation of the entropy functional , we find that the stationary solution of the first - order variation is neither a maximum nor a minimum , but a saddle - point solution .
we argue that such a saddle - point solution by no means indicates a failure of the theory , but rather suggests a new physics that we may have never realized before .
inspired by this saddle - point solution , we distinguish between two types of perturbations in self - gravitating systems , namely , the large - scale mass perturbation and the small - scale density perturbation , which correspond to long - range violent relaxation and short - range two - body relaxation / landau damping , respectively .
these two types of relaxation mechanisms operate in different fashions . for an isolated self - gravitating system ,
the entropy can automatically decrease under long - range violent relaxation , until it terminates and the system attains the equilibrium state with the global minimum entropy .
this result is consistent with antonov s proof , or binney s argument that there are no global maximum entropy states for self - gravitating systems .
however , for every and any small part of the system in its equilibrium state , the entropy is actually the local maximum under short - range relaxations , which is not in conflict at all with the standard second law of thermodynamics .
this minimum - maximum entropy duality constitutes the complementary second law of thermodynamics for long - range self - gravitating systems .
@xcite found the decreasing / increasing property of the general @xmath4-functions in the two - phased violent relaxation , which can be regarded as a support to our finding in this work .
although we have not proposed a complete entropy form , it is extremely intriguing that this simple , or somewhat crude , entropy form of ideal gas , by @xcite , can lead to such a profound new physics .
nevertheless , our study has definitely demonstrated that the concept of entropy is still valid for self - gravitating systems .
however , the standard second law of thermodynamics should be revised , and the term ` principle of maximum entropy ' should be abandoned , and should be simply replaced by ` entropy principle ' .
the validity of entropy for self - gravitating systems means that the statistical mechanics and the relevant thermodynamic relations can also , in principle , be established for such a long - range interaction system .
we believe that our findings , especially the complementary second law of thermodynamics , are interesting , and may offer important clues to the development of statistical mechanics for self - gravitating systems as well as other long - range interaction systems .
ph is very grateful for the comments and suggestions by professor simon d. m. white , drs .
s. hansen , and a. graham .
this work is supported by the chinese academy of sciences under grant no .
kjcx3-syw - n2 , also in part by the project of knowledge innovation program ( pkip ) of chinese academy of sciences under grant no .
kjcx2.yw.w10 and by national basic research program of china , no .
2010cb832805 .
chavanis p. h. , 2002 , in antonic n. , van duijn c.j .
, jager w. , mikelic a. , eds , proceedings of the conference on multiscale problems in science and technology .
springer - verlad , berlin , p. 85
( astro - ph/0212205 ) sylos labini f. , 2008 , in campa a. , giansanti a. , morigi g. , sylos labini f. , eds , dynamics and thermodynamics of systems with long - range interactions : theory and experiments .
melville , new york , p. 205 | the statistical mechanics of isolated collisionless self - gravitating systems is a long - held puzzle , which has not been successfully established for nearly 50 yr . in this work ,
we employ a phenomenological entropy form of ideal gas , first proposed by white & narayan , to revisit this issue . by calculating the first - order variation of the entropy , subject to the usual mass- and energy - conservation constraints
, we obtain an entropy stationary equation .
incorporated with the jeans equation , and by specifying some functional form for the anisotropy parameter @xmath0 , we numerically solve the two equations , and demonstrate that the velocity anisotropy parameter plays an important role to attain a density profile that is finite in mass , energy , and spatial extent . if incorporated again with some empirical density profile from simulations , our theoretical predictions of the anisotropy parameter , and the radial pseudo - phase - space density @xmath1 in the outer non - gravitationally degenerate region of the dark matter halo ,
agree rather well with the simulation data , and the predictions are also acceptable in the middle weak - degenerate region of the dark halo .
the disagreements occur just in the inner strong - degenerate region because of the ignoring of gravitational degeneracy . as far as we know
, our results may be the first theoretical predictions based on the entropy principle that can partially match the empirical data .
the second - order variational calculus reveals the seemingly paradoxical but actually complementary consequence that the equilibrium state of self - gravitating systems is the global minimum entropy state for the whole system under long - range violent relaxation , but simultaneously the local maximum entropy state for every and any small part of the system under short - range two - body relaxation and landau damping .
this minimum - maximum entropy duality means that the standard second law of thermodynamics needs to be re - expressed or generalized for self - gravitating systems .
we believe that our findings , especially the complementary second law of thermodynamics , may provide crucial clues to the development of the statistical physics of self - gravitating systems as well as other long - range interaction systems .
[ firstpage ] methods : analytical cosmology : theory dark matter large - scale structure of universe . |
g. amelino - camelia , _ nature _ * 408 * , 661 ( 2000 ) .
g. amelino - camelia , j. ellis , n. e. mavromatos , d. v. nanopoulos and s. sarkar , _ nature _ * 393 * , 763 ( 1998 ) .
b. e. schaefer , _ phys .
lett . _ * 82 * , 4964 ( 1999 ) .
j. ellis , n. e. mavromatos , d. v. nanopoulos and a. s. sakharov , _ astron .
astrophys . _ * 402 * , 409 ( 2003 ) .
j. ellis et al . , _ astropart .
phys . _ * 25 * , 402 ( 2006 ) .
s. e. boggs , c. b. wunderer , k. hurley and w. coburn , _ apjl _ * 611 * , l77 ( 2004 ) . k. s. cheng and t. harko , _ astropart .
_ * 22 * , 297 ( 2004 ) .
m. gogberashvili , a. s. sakharov and e. k. g. sarkisyan , _ phys .
b _ * 644 * , 179 ( 2007 ) .
r. lamon , n. produit and f. steiner , _
arxiv:0706.4039v2_. j. d. scargle , j. p. norris and j. t. bonnell , _ arxiv : astro - ph/0610571v2_. u. jacob and t. piran , _ nature phys .
_ * 3 * , 87 ( 2007 ) .
m. biesiada and a. pirkowska , _ jcap _ * 0705 * , 011 ( 2007 ) .
j. bolmont et al . , _ arxiv : astro - ph/0603725v6_. m. rodrguez martnez and t. piran , _ jcap _ * 0604 * , 006 ( 2006 ) . | we point out that previous studies of possible lorentz - violating effects in astronomical time - of - flight data did not take into account the entire implications of the universe s cosmological expansion .
we present the derivation of the accurate formulation of the problem and show that the resulting correction of the limits on lorentz violation is significant .
lorentz invariance violation ( liv ) arises in various frameworks and theories of quantum - gravity @xcite . in the context of quantum - gravity
, spacetime may have a non - trivial small - scale structure , so the laws of physics may be altered around the planck scale ( @xmath0 ) . at lower energy scales small deviations from lorentz invariance may be felt . as the effects of liv may become substantial for high - energy particles ,
there has been a growing interest in testing liv by means of time - of - flight measurements of astronomical particles .
time - of - flight analyses provide the most generic method of testing liv .
astronomical sources provide , in addition to high observable energies , large propagation distances that amplify the small liv effects . the use of cosmological sources to gain insight on possible modifications to particles flight times due to liv requires a careful consideration of the universe s cosmological expansion . within the liv phenomenology massless particles
may have energy - dependent speeds , and so high - energy particles may arrive with a delay ( or an early arrival , but we use the term delay here to imply either cases ) compared to low - energy particles emitted at the same instant .
the first attempt to quantify limits on liv by analyzing energy - dependent features in the light curves of gamma - ray bursts ( grbs ) @xcite used an approximation neglecting the cosmological expansion .
later , ellis et al . @xcite performed a systematic liv study based on a combined analysis of an ensemble of grbs at cosmological distances up to @xmath1 .
they were able to place a robust lower bound on the energy scale of linear liv at @xmath2 .
ellis et al .
introduced @xcite a formulation that included the effect of the cosmological expansion , and several groups employed their formulae to perform additional time - of - flight analyses of grbs @xcite .
the different studies produced quantitative limits on the energy scale of liv and also predicted the scales that will be examinable with future experiments . in this short note
we point out a small but important mistake in the calculation of the cosmological liv delays , presented in @xcite and used in the subsequent studies .
we present the derivation of the liv delay expression we introduced in @xcite ( see also @xcite ) and show that this expression entails a significant correction , which changes by a factor of order @xmath3 the liv limits derived earlier .
we consider a general model in which there is a break - down of lorentz symmetry at some very high energy scale ( possibly the quantum - gravity energy scale ) , which we denote as @xmath4 .
when examining particles with energies much smaller than the symmetry breaking scale , we may regard only the leading order correction .
if the leading liv correction is of order @xmath5 , the dispersion relation of massless particles can be generically approximated by : @xmath6 where @xmath7 may be freely substituted with @xmath8 in the liv term , as we are only interested in the leading order correction .
@xmath9 , the conventional speed of light constant , is , in this framework , the speed of low - energy photons .
we choose for clarity to use the @xmath10 case of subluminal motion for the rest of our derivation .
two particles , that are emitted simultaneously from a source and have different propagation speeds , will arrive on earth at different times .
if the source is at a cosmological distance , then as a result of the universe s expansion , the proper ( physical ) distances traveled by the particles will also differ .
a length that is by definition always fixed between the source and the observer ( provided they move together with the universe s expansion ) is the comoving distance. in order to determine the cosmological liv delay between the two particles we have to inspect their comoving trajectories .
the comoving trajectory of a particle is obtained @xcite by writing its hamiltonian in terms of the comoving momentum : @xmath11 where @xmath12 is the cosmological expansion factor .
assuming that the standard relation @xmath13 holds and neglecting higher order liv corrections , we write the comoving path as : @xmath14 where @xmath15 here is a constant , equal to the present - day momentum . rewriting this , we obtain the comoving distance traversed by a massless particle , emitted at redshift @xmath16 and traveling up to redshift @xmath17 : @xmath18 @xmath19 is the redshifted particle energy measured at present .
@xmath20 , @xmath21 and @xmath22 are the cosmological parameters evaluated today .
we examine a low - energy photon , that was emitted at redshift @xmath16 and reaches us at redshift @xmath17 , and a highly energetic one , that was also emitted at redshift @xmath16 and arrives with a delay at redshift @xmath23 .
the comoving distance , traveled by both particles , emitted from the same source and reaching earth , is the same . equating the two paths and taking again only the leading order liv corrections yields : @xmath24 since for all delays
that may be considered @xmath25 is a very small number , we neglect second order corrections in @xmath25 and arrive at the expression for the liv time delay of a cosmological high - energy massless particle : @xmath26 it is worth noting that the same expression is valid for an ultrarelativistic massive particle , such as an energetic neutrino @xcite , if the correction term in the dispersion relation due to the particle s mass is smaller than the correction due to liv
. in @xcite ellis et al . derived
the delay between two photons arriving from redshift @xmath16 with present - day energy difference @xmath27 for the case of a linear liv induced correction to the dispersion relation .
their result in our notations , @xmath28 , lacks a factor of @xmath3 in comparison with our expression ( for @xmath29 ) .
the reason for this is that the derivation in @xcite equates the proper distances that the two photons traverse by the time they reach earth .
as mentioned , this is incorrect - the source and the earth have a fixed comoving distance between them , but the proper distance varies as the universe expands . while the energetic photon is delayed throughout the path , the universe s expansion factor is very slightly modified , but this slight modification affects the large cosmological distance that the particle must pass before reaching the observer .
hence , this effect accumulates to give a significant difference in the proper distances traveled .
a detailed calculation shows that the time delay due to the additional proper distance that the slower particle passes is of the same order of magnitude as the delay produced between the particles by the time they complete the same proper distance .
thus , when inspecting the delays of arrival on earth of particles propagating from distant cosmological sources , the expansion of the universe during the delay periods can not be neglected .
another simple illustration of the problem with the derivation in @xcite is that the delays produced between the particles at different intervals along the path are merely summed up , without considering the fact that a delay produced further back in the path amounts to a larger delay on earth .
this effect of relativistic dilation introduces a factor of @xmath3 into the above integral .
although we are dealing with tiny modifications to conventional physics that produce short delays , the detailed cosmological analysis is necessary and the correction we presented here is significant . for example , using standard cosmology ( @xmath30 , @xmath31 , @xmath32 ) and referring to the scenario of @xmath29 liv with @xmath33 ( the symmetry breaking scale is the planck energy ) , we obtain with ( [ liddelay ] ) for a @xmath34 photon emitted at @xmath35 a delay of @xmath36 , in comparison to a delay of @xmath37 according to the calculation in @xcite . for more distant sources the factor between the correct calculation and the erroneous
one will be even larger .
we find that the required correction of the cosmological liv delay calculations would serve to extract stronger liv bounds from the previously analyzed grb data .
for example , given the range of redshifts in the sample of sources used by ellis et al . , this correction implies that the limits are stronger by a factor of @xmath38 . |
it is widely accepted by some time that merging and interaction events play an important role in the formation and evolution of galaxies .
mergers change the mass function of galaxies , creating a progression from small galaxies to larger ones ; the merging process can also change the morphology of the constituents , transforming gas - rich spirals in quiescent ellipticals .
interactions can also trigger a wide set of physical and morphological phenomena , such as tidal tails , bridges and shells , kinematically decoupled cores and star formation enhancements ( see review of @xcite ) .
interacting galaxies also show enhanced star formation when compared with isolated galaxies .
such enhancement was initially proposed by @xcite to explain the wider range of optical colors found in galaxies in pairs . since then , numerous studies have confirmed these results , especially in the central regions , through measurements of optical emission lines @xcite , infrared emission @xcite and radio emission @xcite .
recent studies have also shown that this enhancement is a function of the projected galaxy pair separation ( e.g. @xcite ) , being stronger in low - mass than in high - mass galaxies ( e.g. @xcite ) .
in particular , @xcite found that the star formation rate as measured by the h@xmath2 equivalent width for galaxies in pairs selected from the sloan digital sky survey , is some 70% higher when compared to a control sample of galaxies with equal stellar mass distribution .
there is a connection between the interaction strength and the morphological distortion in binary galaxies . according to @xcite models , the response of the gas to a close passage depends dramatically on the mass distribution of the galaxy , with the irregularities in the gas velocity field tracing the disturbances in the gravitational potential of the galaxy as observed , for example , in some galaxies in the virgo cluster @xcite .
combined n - body / hydrodynamic simulations show that galaxy - galaxy mergers disturb the gas velocity field significantly , and hence lead to asymmetries and distortions in the rotation curves @xcite .
however , according to these authors , no severe distortions are observable about 1 gyr after the first encounter .
the gas motions created by the interaction can also significantly alter the chemical state of the galaxies @xcite , and modify the usually smooth radial metallicity gradient often found in isolated disk galaxies @xcite .
recently , @xcite found that o / h abundance in the central region of nearby galaxy pairs is systematically lower that that of isolated objects .
these authors suggest that the lower metallicity is a consequence of gas infall caused by the interaction , but very few observational studies have been published analysing in detail the influence of different levels of interactions in the metallicity distribution and enrichment properties of galaxies .
ferreiro & pastoriza 2004 ( hereafter fp04 ) , in a study of the integrated photometry and star formation activity in a sample of interacting systems from the arp - madore catalogue @xcite , found that the galaxies involved have bluer colours than those of isolated galaxies of the same morphological type , indicating an enhancement of star forming activity .
this enhancement was also previously suggested by @xcite to explain the slightly larger values of h@xmath3n@xmath4ii equivalent widths found in these systems , when compared with normal isolated spiral galaxies . from their sample
, we have selected several systems to start a more comprehensive study of the effects of the interactions in the kinematics , stellar population and abundances of the galaxies in the so - called `` minor merger '' systems , defined here as physical pairs with mass ratio in the range of @xmath5 .
this paper presents the results for the system am2306 - 721 .
this pair is composed by a peculiar spiral with disturbed arms ( hereafter , am2306a ) in interaction with an irregular galaxy ( hereafter , am2306b ) .
both galaxies contain very luminous h@xmath4ii regions with h@xmath2 luminosity in the range of @xmath6l(h@xmath7erg s@xmath0 as estimated from h@xmath8 images @xcite ; and high star formation rate in the range of 0.07 to 10 @xmath9/yr @xcite .
the present paper is organized as follows : in section [ datared ] , we summarize the observations and data reduction . section [ vel ] describes the gas kinematics of each galaxy and section [ numsim ] present the numerical n - body / hydrodynamical simulations of the interaction .
section [ sintese ] presents the stellar population synthesis .
the metallicity analysis is presented in section [ emission ] , and the conclusions are summarized in section [ final ] .
long slit spectroscopic data was obtained in 20/21 june 2006 and 23/24 june 2007 with the gemini multi - object spectrograph at gemini south , as part of poor weather programmes gs-2006a - dd-6 and gs-2007a - q-76 .
spectra in the range 3350 to 7130 were taken in two settings with the b600 grating , and the 1@xmath10 slit , yielding a spectral resolution of 5.5 .
the frames were binned on - chip by 4 and 2 pixels in the spatial and spectra directions , respectively , resulting in a spatial scale of 0.288 @xmath10pxl@xmath0 , and 0.9pxl@xmath0 dispersion .
spectra were taken at three different position angles on the sky : pa=238@xmath11 and pa=118@xmath11 corresponding to positions along the major axis of am2306a and am2306b , respectively ; and pa=190@xmath11 , a position cutting across the disk of both galaxies .
the exposure time on each single frame was limited to 700 seconds to minimize the effects of cosmic rays , with four frames being obtained for each slit position to achieve suitable signal .
the slit positions are shown in fig .
[ field ] , superimposed on the @xmath12-band image of the pair .
table [ observ ] gives the journal of observations .
conditions during both runs were not photometric , with thin cirrus and image quality in the range 1.0@xmath10 to 2.5@xmath10 ( as measured from stars in the acquisition images taken just prior to the spectroscopic observations ) .
the spectroscopic data reduction was carried out using the gemini.gmos package as well as generic iraf tasks .
we followed the standard procedure for bias correction , flat - fielding , cosmic ray cleaning , sky subtraction , wavelength and relative flux calibrations . in order to increase the signal - to - noise ratio ,
the spectra were then extracted summing over six rows .
each spectrum thus represents an aperture of 1@xmath13 . for a distance of 119mpc for am2306a , and 116mpc for am2306b , ( estimated from the radial velocities derived in section [ vel ] and adopting @xmath14=75kms@xmath0mpc@xmath0 ) , this aperture corresponds to a region of 577@xmath15998pc@xmath16 for am2306a ; and 562@xmath15972pc@xmath16 for am2306b .
the nominal centre of each slit was chosen to be the continuum peak at @xmath17 .
[ spectraa ] and [ spectrab ] shows a sample of spectra of am2306a and am2306b extracted along the slit for pa=238@xmath11 and pa=118@xmath11 , respectively .
-band image of am2306 - 721 with the observed slit positions . ] lccc date ( ut ) & exposure time(s ) & pa ( ) & @xmath18 + 2006/06/20 & 4@xmath15700 & 190 & 4280 - 7130 + 2006/06/20 & 4@xmath15700 & 238 & 4280 - 7130 + 2007/06/24 & 4@xmath15700 & 190 & 3350 - 6090 + 2007/06/24 & 4@xmath15700 & 238 & 3350 - 6090 + 2007/06/24 & 4@xmath15700 & 118 & 3350 - 6090 +
the radial velocity was estimated from the strongest emission lines present in the spectra , namely @xmath19 @xmath20 , @xmath21 @xmath22,[o@xmath4iii ] @xmath23 , @xmath24 @xmath25 , [ n@xmath4ii ] @xmath26 , and [ s@xmath4ii ] @xmath27 .
the final radial velocity for each spectrum was obtained by averaging the individual measurements from the detected emission lines , and the errors estimated from the standard deviation of these measurements around that mean . using the acquisition images in the @xmath12 filter
, we did a simple isophote fitting using the stsdas.ellipse task to obtain the position angle of the major and minor axes of both galaxies .
we found that the major axis for am2306a is at pa=@xmath28 , and for am2306b at pa=@xmath29 .
the inclination of each galaxy with respect to the plane of the sky was also computed as @xmath30 , where @xmath31 and @xmath32 are the minor and major semi - axes of the galaxy , respectively .
we obtained @xmath33 for the main galaxy and @xmath34 for the secondary , in agreement with the values measured by fp04 .
the rotation curves along the observed slit positions for both galaxies are presented in figure [ radial_velocity ] , as well as the coresponding two - dimentsional spectra of h@xmath2 and/or h@xmath35 emission lines .
the radial velocity values after subtraction of the systemic velocities ( as determined from the rotation model described below ) are listed in table [ velo ] .
both galaxies show a fairly symmetric rotation curve , so we adopted a very simple approximation for the observed velocity distribution , assuming that the gas moves under a logarithmic gravitational potential , following circular orbits close to a plane @xmath36 , characterized by its inclination to the plane of the sky @xmath37 and the position angle ( pa ) of the line of nodes @xmath38 .
this assumption results in an observed radial circular velocity @xmath39 in the plane of the sky given by @xcite : @xmath40 with @xmath41 , \label{v_molcont}\ ] ] where @xmath42 is the systemic velocity , @xmath43 is the radius in the plane of the galaxy and @xmath44 and @xmath45 are parameters that define the amplitude and shape of the curve .
the fit of the rotation curves for each galaxy are shown in fig .
[ radial_velocity ] and the parameters obtained are listed in table [ par ] .
lccc & & & + parameter & model & & model + @xmath42 ( km / s ) & 8919@xmath465 & & 8669@xmath463 + @xmath44 ( km / s ) & 175@xmath47 & & 185 @xmath466 + @xmath45 ( kpc ) & [email protected] & & 1.6 @xmath460.2 + [ velo ] rrrrrrrrrrr & & + & & & & & & + & & & & & & & & & & + 13.96 sw & 204@xmath4610 & & 13.96 s & ... & & ... & ... & & ... & ... + 12.96 sw & 131@xmath468 & & 12.96 s & ... & & ... & ... & & ... & ... + 11.96 sw & 150@xmath465 & & 11.96 s & ... & & ... & ... & & ... & ... + 10.97 sw & 118@xmath4616 & & 10.97 s & ... & & ... & ... & & ... & ... + 9.97 sw & 134@xmath4616 & & 9.97 s & 125@xmath4618 & & ... & ... & & ... & ... + l c c & * am2306a * & * am2306b * + number of points in disk & 2048 & 2048 + disk mass & 0.5 & 0.17 + disk radial scale length & 0.6 & 0.6 + disk vertical scale thickness & 0.06 & 0.05 + reference radius r@xmath48 & 0.06 & 1.2 + toomre q at r@xmath48 & 1.5 & 1.5 + number of points in gas disk & 2048 & 2048 + gas disk mass & 0.05 & 0.017 + gas disk radial scale length & 1.0 & 0.75 + gas disk vertical scale thickness & 0.02 & 0.03 + toomre q at r@xmath48 & 1.5 & 1.5 + number of points in bulge & 256 & 256 + bulge mass & 0.08 & 0.01 + bulge radial scale length & 0.1 & 0.06 + number of points in spherical halo & 2048 & 2048 + halo mass & 3.5 & 1.7 + halo cutoff radius & 8.0 & 4.0 + halo core radius & 2.5 & 0.6 + * notes : * simulations were done in a system of units with g=1 .
model units scale to physical ones as : unit length is 3.5kpc , unit velocity is 262kms@xmath0 , unit mass is @xmath49 and unit time is 13.062myr .
the heliocentric velocity of the main galaxy is found to be 8919 km s@xmath0 .
the rotation curve obtained for am2306a along pa=238 is well represented by the simple model above .
this rotation curve is typical of spiral disks , rising shallowly and flattening at an observed amplitude of 145 kms@xmath0 . as can be seen in fig.[radial_velocity ] ,
there are pronounced velocity features indicating local deviations from a smooth rotational field .
one of such well defined features is located towards the northeast along pa=238 , at about 8 kpc projected distance from the nucleus , with an amplitude ( peak - to - peak ) of about 100 kms@xmath0 .
this feature is located where the slit crosses the the outer spiral arm as seen in the @xmath50 image ( fig .
[ haimage ] , right ) .
this image also makes more obvious the disturbed spiral structure of the main component , very likely due to the interaction with am2306b . for the secondary galaxy , the rotation curve model results in an heliocentric velocity of 8669 kms@xmath0 .
the observed radial velocities along pa=118 are also very well represented by the model .
the rotation curve is quite similar to that of am2306a , rising shallowly and flattening at an observed amplitude of about 158 km s@xmath0 .
although the young star forming population of the secondary galaxy show clear morphological irregularities ( fig .
[ haimage ] , left ) , the rotation curve does not present significant deviations from the smooth representation of the velocity field .
@xcite estimated systemic velocities of 8762 and 9069 kms@xmath0 for am2306a and am2306b , respectively .
following our results , we verified that the values presented by those authors were reversed , and therefore our estimates are found to agree to within 2 % with those previous determinations .
an estimate of the dynamical mass can be derived by assuming that the mass inside a certain radius is given by @xmath51 . for the main galaxy ,
the deprojected velocity amplitude is 175 km s@xmath0 and its dynamical mass is @xmath52 within a radius of 18 kpc . for the companion galaxy ,
the deprojected velocity amplitude is 185 km s@xmath0 and the dynamical mass within a radius of 10.7 kpc is @xmath53 .
it is important to emphasize that the maximum radius to which we can observe the gas in emission is quite certainly smaller than the total radius of the galaxies , so our estimates of the dynamical mass give a lower limit to the actual dynamical mass of each system . in determining the dynamical mass ,
the main source of error is in the estimation of the deprojected velocity , which in turn is highly dependent on the assumed inclination of the galaxies with respect to the plane of the sky .
for am2306a the inclination angle is well determined from the isophote fitting , with an estimated error of about 1% , propagating to a 2% uncertainty in the resulting total mass . for am2306b
, the irregular morphology of the disk in optical images implies that the calculated value of the inclination angle varies significantly with radius . in this case , if we considered the isophote fitting from the regions more perturbed of am2306b to estimate the inclination angle , a maximum angle inclination of about 12 % would be obtained , implying in a reduction of up to 18 % in the total mass .
in order to reconstruct the history of the am2306 - 721 system and to predict the evolution of the encounter , we modeled the interaction between am2306a and am2306b through numerical simulations using the the n - body / sph code gadget-2 developed by @xcite .
the galaxies were modeled following the prescription of @xcite , including a gaseous disk component . the observed morphology and rotation curves presented in sec .
[ vel ] provide the constraints to the simulations . as for any study of this kind
, recreating the evolution of the am2306 - 721 system requires solving the reverse problem of calculating the orbit followed by the two galaxies from their observed properties .
this , like most similar cases , is not a fully determined problem , since the observations do not provide all the necessary information to uniquely identify the solution .
therefore , in order to restrict the parameter space when setting up the initial conditions for the simulations , we first calculate orbits that satisfy the constraint given by the observed radial velocity difference , testing different eccentricities , pericenter and line - of - sight direction distances . from that subset , based on the observed morphology , we select a few orbits to run the full simulations , from which the one that best fits the observed properties is selected .
the orbit that best reproduces the observational properties is found to be hyperbolic , with an eccentricity @xmath54 and perigalacticum of @xmath55 kpc .
the orbital plane is perpendicular to the plane of the sky , and intersects the later in the line that connects am2306a and am2306b as projected in the sky . in perigalacticum , am2306b was behind am2306a , along the line of sight .
the parameters for the best fit model are presented in table [ tab : modelpars ] .
the composite rotation curve , as well as the individual model components and observed circular velocity , corrected by the inclination of each galaxy as given in section [ vel ] , are shown in figure [ vcirc ] .
the models indicate a total mass of @xmath56 for am2306a and of @xmath57 for am2306b .
the mass of each individual component ( disk , bulge , halo and gas ) is quoted in table [ tab : modelpars ] .
figure [ simul ] shows the time evolution of the encounter .
simulation starts 360 myr before perigalacticum , and the time is shown in myr in the upper right corner of each frame .
the situation that best reproduces the morphology and kinematics of am2306 - 721 system at the current stage is @xmath58 myr , or about 250 myr after perigalacticum .
the overall large scale morphology and kinematics agree well with observations , considering the low resolution of the simulation .
a detailed study of the star formation in minor merger galaxies is an important source of information not only on the age distribution of their stellar population components , but to better understand several aspects related to the interacting process , its effect in the properties of the individual galaxies and their later evolution .
the absorption features arising from the stellar component also affect to different degrees the measured intensity of the emission line in the spectrum of the gaseous component .
this effect is more prominent in , but not restricted to , the balmer lines , so the stellar population contribution must be subtracted from each spectra in order to investigate the physical properties of the gas in these galaxies . to investigate the star formation history of am2306a and am2306b we use the stellar population synthesis method developed by @xcite .
this method employs the equivalent widths @xmath59 of several spectral absorption features and the measured continuum fluxes @xmath60 at different wavelengths , comparing them to those of a model computed from a base of simple stellar population ( ssp ) elements with known ages and metallicities .
the algorithm generates all possible combinations of the base elements according to a given flux contribution step and compares the resulting @xmath59 and continuum points to the input ones .
the allowed solutions are those which reproduce , within predefined limits , the observed @xmath59 and @xmath60 .
all these possible solutions are then averaged , and this average is adopted as the final synthesis , with the uncertainty for each individual age component given by the standard deviation of the contribution from each allowed solution around the mean adopted value .
the code used here is based on an upgraded version of the one presented by @xcite , and includes a correction for the effect of the internal extinction . the spectral windows for measuring @xmath59 are those defined in the lick system @xcite , with the addition of the k ca@xmath4ii ( corresponding to the spectral window @xmath61 ) ) , and h11 @xmath62 lines ( spectral window @xmath63 ) .
previous to the measurements of the @xmath59 and @xmath60 the spectra were corrected by the foreground ( galactic ) reddening of @xmath64 mag taken from @xcite ; and normalized to @xmath65 .
the @xmath59 were measured using the _ pacce _ code kindly provided by @xcite .
the @xmath59 values measured for each galaxy at different galactocentric distances and position angles are listed in table [ eqw ] .
the ssp base was created from the @xcite evolutionary stellar population models , which are based on a high resolution library of observed stellar spectra .
this library allows us to derive detailed spectral evolution of simple stellar populations across the wavelength range of 3200 to 9500 @xmath66 with a wide range of metallicities .
we used the padova 1994 tracks as recommended by @xcite , with the initial mass function of salpeter @xcite .
the final base contains five spectra corresponding to ssps with ages of 2.5 myr , 200 myr , 1 gyr , 5 gyr and 10 gyr , and solar metallicity .
the adopted number of elements in the base was defined by a compromise between having enough age resolution , having a limited number of features available to constrain the fit , and the desire to mantain a consistent base for all apertures in both galaxies .
for the few positions with higher s / n spectra in each galaxy , we tested a more detailed age grid , including components with ages 2.5 myr , 100 myr , 200 myr , 500 myr 1 gyr , 5 gyr and 10 gyr , and solar metallicity .
all those converged to the same results obtained with the original five component base , with the contributions from the 100 and 500 myr populations resulting always less than 5 % .
figure [ sintesec ] show an example of the observed spectra corrected for the reddening , the synthesized spectra and the pure emission spectra .
table [ sint_result ] presents the result of the stellar population synthesis for the individual spatial bins in each galaxy , stated as the percentual contribution of each base element to the flux at @xmath67 .
the values of @xmath68 derived from the synthesis are also given in table [ sint_result ] .
lccccccc & + r(kpc ) & @xmath69 & @xmath70&g @xmath71 & fe@xmath4i @xmath72 & fe@xmath4i @xmath73 & fe@xmath4i @xmath74 & nad @xmath75 + + 3.99 sw & [email protected] & ... & ... & ... & ... & [email protected] & [email protected] + 2.99 sw & [email protected] & [email protected] & ... & [email protected] & ... & [email protected] & [email protected] + 1.99 sw & ... & [email protected] & [email protected] & [email protected] & ... & [email protected] & [email protected] + 1.00 sw & [email protected] & [email protected] & [email protected] & [email protected] & ... & [email protected] & [email protected] + 0 & ... & [email protected] & [email protected] & [email protected] & ... & [email protected] & [email protected] + the spatial variations in the contributions of the stellar population components are shown in fig .
[ pop_c ] .
large variations in the contribution from the different age components can be seen across the disk of both galaxies .
the central regions of am2306a are dominated by the old ( 5 - 10gyr ) population , with some significant contribution from a young 200myr and intermediate 1gyr component of the spiral arms and the disk . on the other hand ,
the stellar population in am2306b is overall much younger , being dominated by the 2.5myr , 200myr and 1gyr components , which are quite widely spread over the whole disk .
the star formation episode occurred about 200myr ago in both galaxies could be related with the perigalactic passage .
lrrrrrrc & + & & & & & & + + 3.99 sw & 15@xmath461 & 52@xmath462 & 3@xmath4612 & 15@xmath4662 & 15@xmath4643 & [email protected] + 2.99 sw & 2@xmath461 & 14@xmath461 & 77@xmath469 & 1@xmath461 & 6@xmath4610 & [email protected] + 1.99 sw & 2@xmath461 & 1@xmath461 & 76@xmath4614 & 13@xmath4642 & 8@xmath4625 & [email protected] + 1.00 sw & 2@xmath461 & 8@xmath461 & 29@xmath464 & 16@xmath4646 & 45@xmath4628 & [email protected] + 0 & 2@xmath461 & 22@xmath461 & 7@xmath469 & 13@xmath4672 & 56@xmath4647 & [email protected] + 1.00 ne & 1@xmath461 & 23@xmath463 & 1@xmath461 & 5@xmath4627 & 70@xmath4624 & [email protected] + 1.99 ne & 3@xmath461 & 1@xmath462 & 79@xmath468 & 12@xmath4622 & 5@xmath4613 & [email protected] + 2.99 ne & 3@xmath461 & 21@xmath462 & 75@xmath461 & 1@xmath461 & 0@xmath461 & [email protected] + 3.99 ne & 2@xmath461 & 24@xmath466 & 72@xmath469 & 1@xmath462 & 1@xmath461 & [email protected] + 4.98 ne & 0@xmath461 & 30@xmath461 & 49@xmath4611 & 7@xmath4619 & 14@xmath462 & [email protected] + 5.98 ne & 0@xmath461 & 39@xmath462 & 54@xmath4618 & 4@xmath4611 & 3@xmath467 & [email protected] + 6.98 ne & 0@xmath461 & 35@xmath466 & 63@xmath4610 & 1@xmath462 & 1@xmath461 & [email protected] + 7.98 ne & 0@xmath461 & 50@xmath461 & 31@xmath4610 & 9@xmath4632 & 10@xmath4627 & [email protected] + + 2.91 se & 32@xmath463 & 67@xmath466 & 1@xmath461 & 0@xmath460 & 0@xmath460 & [email protected] + 1.94 se & 43@xmath463 & 55@xmath465 & 0@xmath461 & 1@xmath461 & 1@xmath461 & [email protected] + 0.97 se & 22@xmath461 & 53@xmath462 & 11@xmath468 & 7@xmath4618 & 7@xmath4611 & [email protected] + 0 & 16@xmath462 & 47@xmath464 & 13@xmath4614&14@xmath4654 & 10@xmath4630 & [email protected] + 0.97 nw & 21@xmath461 & 45@xmath462 & 30@xmath466 & 2@xmath465 & 2@xmath463 & [email protected] + 1.94 nw & 20@xmath461 & 41@xmath462 & 36@xmath466 & 2@xmath465 & 1@xmath463 & [email protected] + 2.91 nw & 19@xmath461 & 42@xmath462 & 37@xmath463 & 1@xmath463 & 1@xmath462 & [email protected] + 3.88 nw & 15@xmath461 & 44@xmath462 & 36@xmath464 & 3@xmath464 & 2@xmath463 & [email protected] + 4.86 nw & 35@xmath461 & 63@xmath463 & 1@xmath461 & 1@xmath461 & 0@xmath461 & [email protected] + 5.83 nw & 19@xmath461 & 44@xmath461 & 34@xmath466 & 2@xmath465 & 1@xmath464 & [email protected] + 6.80 nw & 23@xmath461 & 25@xmath461 & 47@xmath4620 & 3@xmath4611 & 2@xmath464 & [email protected] + 7.77 nw & 16@xmath461 & 46@xmath462 & 36@xmath465 & 1@xmath464 & 1@xmath462 & [email protected] +
once the stellar population contribution has been determined , the underlying absorption line spectrum can be subtracted to allow the correct measurement and analysis of the line emission from the gaseous component .
the line intensities were measured using gaussian line profile fitting on the pure emission spectra .
we used the iraf splot routine to fit the lines , with the associated error being given as @xmath76 , where @xmath77 and @xmath78 are the continuum rms and the poisson error of the line flux , respectively .
the residual extinction associated with the gaseous component for each spatial bin was calculated comparing the observed @xmath79 and @xmath80 ratios to the theoretical values in @xcite for an electron temperature of 10000 @xmath81 and a density of 100 @xmath82 .
the observed emission line intensities were then corrected by this residual extinction using the @xcite reddening function .
table [ lines ] present the main emission line intensities normalized to the flux in the @xmath21 line .
we derived the electron densities for am2306a from the [ s@xmath4ii]@xmath83 intensity ratios , using the _ temden _ routine of the _ nebular _ package from _ stsdas / iraf _ , assuming an electron temperature of 10000 k. the energy levels , transition probabilities and collisional strength values for [ s@xmath4ii ] were taken respectively from @xcite , @xcite and @xcite .
the resulting electron density obtained for am2306a is in the low density limit for the [ s@xmath4ii ] lines , with values below @xmath84 , compatible with the results for both galactic @xcite , and giant extragalactic h@xmath4ii regions @xcite .
[ lines ] rrrrrrrrrrrr & + & c(h@xmath35 ) & & & & & & & & & & + 12.96 sw & 0 & ... & ... & 100@xmath463 & 40@xmath462 & 118@xmath464 & 22@xmath464 & 286@xmath4619 & 60@xmath465 & 73@xmath4610 & 73@xmath4610 + 11.96 sw & 0.10 & ... & 47@xmath469 & 100@xmath464 & ... & 72@xmath464 & 29@xmath467 & 286@xmath4625 & 94@xmath4610 & 94@xmath4614 & 86@xmath4614 + 10.97 sw & 0 & ... & ... & 100@xmath465 & ... & 62@xmath464 & 54@xmath4613 & 286@xmath4633 & 117@xmath4617 & ... & ... + 9.97 sw & 0 & ... & ... & 100@xmath464 & ... & 55@xmath463 & ... & 286@xmath4627 & 121@xmath4615 & ... & ... + 8.97 sw & 0 & ... & ... & 100@xmath463 & ... & 61@xmath463 & 34@xmath465 & 286@xmath4618 & 101@xmath468 & 89@xmath464 & 65@xmath464 + the spatial metallicity variations across the disk of spiral galaxies is a fundamental parameter for understanding galaxy evolution . in isolated spiral galaxies
, it is often found that the metallicity in the ionized interstellar medium decreases outwards @xcite , while a weaker or absent metallicity gradient in interacting galaxies of similar morphological type have been pointed out as evidence of radial gas inflows driven by tidal torques generated during the interaction @xcite .
determining accurate element abundances from optical spectra is critically dependent on measuring temperature sensitive line ratios , such as [ o@xmath4iii]@xmath85 .
however , when doing spectroscopy of h@xmath4ii regions with high metallicity and/or low excitation , temperature sensitive lines such as [ o@xmath4iii]@xmath86 are found to be weak or unobservable , and empirical indicators based on more easily measured line ratios have to be used to estimate metal abundances . the line ratio @xmath1=([o@xmath4ii]@xmath87[o@xmath4iii ] @xmath88[o@xmath4iii ]
@xmath89h@xmath35 , introduced by @xcite has been widely used for determination of oxygen abundance in h@xmath4ii regions @xcite .
although the @xmath1 indicator is sensitive to abundance variations , it is double valued , presenting high to low metallicity turnover in the range of @xmath90 @xcite .
another critical issue regarding the use of the @xmath1 is the fact that it is also sensitive to the ionization level of the emission gas , especially at low metallicities .
these problems can be solved , by choosing either the low or high - metallicity branch of @xmath1 , and combining it with another line indicator sensitive to the ionization level , such as the ratio [ o@xmath4ii](@xmath91[o@xmath4iii]@xmath92 @xcite .
since we do not detect any temperature - sensitive emission lines in our spectra , we used the @xmath1 indicator to estimate the metallicity comparing the observed values with a grid of photoionization models obtained from the code cloudy/96.03 @xcite .
the grid was built following the same procedures as @xcite , with metallicities of @xmath93 2.0 , 1.1 , 1.0 , 0.6 , and 0.4 @xmath94 , and ionization parameter @xmath95 @xmath96 , @xmath97 , and @xmath98 .
the solar value of @xmath99 is taken from @xcite .
for each model , the ionizing source was assumed to be a stellar cluster with energy distribution obtained using the stellar population synthesis code @xmath100 @xcite , with an upper mass limit of 100 @xmath101 and an age of 2.5 myr .
figure [ grade ] shows the @xmath1 versus [ o@xmath4ii]/[o@xmath4iii ] diagramme , with the observed values superposed in the computed models .
open squares correspond to regions in am2306a , filled triangles to regions in am2306b .
most of the points indicate gas with solar or less than solar abundance , with the exception of two spatial bins in am2306a , corresponding to the nuclear and the 1.0kpc ne regions .
figures [ abund_c ] and [ abund_b ] shows the estimated o / h abundance distribution as a function of the galactocentric distance , @xmath102 , for am2306a and am2306b , respectively .
the disk of the main galaxy ( fig . [ abund_c ] ) shows a clear radial oxygen abundance gradient , that can be fitted as a linear function @xmath103 , where @xmath102 is in kpc .
this abundance gradient is typical of spiral galaxies , as found in m101 @xcite , and in ngc1365 @xcite .
for am2306b the oxygen abundance is relatively homogeneous across the galaxy disk , not presenting a radial oxygen abundance gradient .
the observed values are in the range @xmath104 .
the above results are in agreement with the luminosity - metallicity relation for gas rich galaxies ( e.g. @xcite ) , in the sense that more luminous objects are more metallic . to investigate if the interaction affected the gas metallicity in both galaxies , we compare the central o / h value of am2306a and am2306b with isolated and pair galaxies from the sample of @xcite .
we found the central o / h value for am2306a is in agreement with the results obtained by @xcite for field galaxies at same luminosity , indicating that the interaction has not significantly affected the gas abundance in am2306a . on the other hand ,
the central o / h value obtained for am2306b is about 0.1 - 0.2 dex lower than those estimated by @xcite for field galaxies at same luminosity , but is in good agreement with the values obtained for galaxies in pairs .
interestingly , the mean value for am2306b , @xmath105 , is similar to the values found in the outer parts of am2306a .
if we assume that in am2306b there was an oxygen abundance gradient before the encounter with am2306a , and that this gradient was destroyed by gas flows from the outer parts to the center of the galaxy , mixing and homogenizing the chemical composition of the interstellar medium @xcite , we can estimate the fraction of infalling gas that was required to produce the metallicities that are now observed in am2306b .
for this , we assume that , before the interaction , am2306b presented the same o / h gradient observed now in am2306a . to reduce the initial abundance of @xmath106 in the central region to the current mean value of 8.39 , it would be necessary that the central gas was diluted so as to contain a fraction of 56% of poor gas from the outer disk plus 44% of rich gas from the central region of the galaxy . another possible explanation to mix the interstellar medium and flatten the radial gradient
would be the presence of a bar @xcite .
however , the very disturbed luminosity profile of this galaxy in @xmath107 , @xmath108 and @xmath109 images ( fp04 ) do not show any evidence of such feature . we can estimate the amount of ionized gas associated with the central star - forming complex in am2306b from the h@xmath2 luminosity of this region as given by @xcite . for @xmath110kpc
, we have @xmath111ergs@xmath0 ( fp04 ) , and assuming a a filling factor of @xmath112 , we find a mass of ionized gas of @xmath113 . according to the above proposed scenario , 56% of this total mass ( @xmath114 ) would be infalling gas from the outer parts of the galaxy .
since we do not see any evidence of ionized gas flows in the radial velocities , we have to assume that the process of gas infall has already stopped .
considering the result of the numerical simulations , we take a timescale of @xmath115100myr for the infall process ( the gas motions were triggered by the perigalactic passage 250myr ago , but are no longer observable ) , resulting in an infall rate of @xmath116/yr .
this infall rate is considerably smaller than the average gas inflow rate of @xmath117/yr to the central 1 - 2 kpc predicted by recent merger models @xcite .
an observational study about the effects of the interactions in the kinematics , stellar population and abundances of the galaxy pair am2306 - 721 is performed .
the data consist of long - slit spectra in the wavelength range of 3350 to 7130 obtained with the gemini multi - object spectrograph at gemini south .
the main findings are the following : 1 .
rotation curves of the main and companion galaxies with an deprojected velocity amplitude of 175 km s@xmath0 and 185 km s@xmath0 respectively were obtained .
an estimate of the dynamical mass was derived for each galaxy , using the deprojected velocity amplitude . for the main galaxy ,
its dynamical mass is @xmath52 within a radius of 18 kpc ; and for the companion galaxy , the estimated dynamical mass is @xmath53 within a radius of 10.7 kpc .
2 . in the main galaxy , radial velocity deviations from the disk rotation of about 100 kms@xmath0 were detected , which are probably due to the interaction with the companion galaxy .
3 . in order to reconstruct the history of the am2306 - 721 system and to predict the evolution of the encounter , we modeled the interaction between am2306a and am2306b through numerical n - body / hydrodynamical simulations .
the orbit that best reproduces the observational properties is found to be hyperbolic , with an eccentricity @xmath54 and perigalacticum of @xmath55 kpc ; the current stage of the system would be about 250 myr after perigalacticum .
the spatial variations of the stellar population components of the galaxies were analysed by fitting combinations of stellar population models of different ages ( 2.5 myr , 200 myr , 1 gyr , 5 gyr and 10 gyr ) and solar metallicity .
the central regions of the main galaxy are dominated by the old ( 5 - 10gyr ) population , with some significant contribution from a young 200myr and intermediate 1gyr component along the disk of the galaxy . on the other hand ,
the stellar population in the companion galaxy is overall much younger , being dominated by the 2.5myr , 200myr and 1gyr components , which are quite widely spread over the whole disk .
oxygen abundance spatial profiles were obtained using a grid of photoionization models and the @xmath1=([o@xmath4ii]@xmath87[o@xmath4iii ] @xmath88[o@xmath4iii ] @xmath89h@xmath35 line ratio .
the disk of the main galaxy shows a clear radial oxygen abundance gradient , that can be fitted as a linear function @xmath118 ; in the companion galaxy the oxygen abundance is relatively homogeneous across the galaxy disk , with the observed mean value of @xmath105 6 .
the absence of an abundance gradient in the secondary galaxy is interpreted as it having been destroyed by gas flows from the outer parts to the center of the galaxy . based on observations obtained at the gemini observatory , which 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 ) , ministrio da ciencia e tecnologia ( brazil ) and secyt ( argentina ) we thank o. l. dors , r.riffel , r.r.riffel and c. bonatto for helpful discussions .
the authors would like to thank volker springel for providing them with gadget-2 .
this research has been partially supported by the brazilian institution cnpq and pronex / fapergs(05/11476 ) .
allende prieto , c. , lambert , d. l. , & asplund , m. 2001 , apj , 556 , l63 arp , h. c. , & madore , b. f. 1987 , a catalogue of southern peculiar galaxies and associations 2 volume set , by halton c. arp and barry f. madore , pp .
isbn 0521343364 .
cambridge , uk : cambridge university press , may 1987 barnes , j. e. , & hernquist , l. 1996 , apj , 471 , 115 barton , e. j. , geller , m. j. , & kenyon , s. j. 2000 , apj , 530 , 660 barton gillespie , e. , geller , m. j. , & kenyon , s. j. 2003 , apj , 582 , 668 bertola , f. , bettoni , d. , danziger , j. , sadler , e. , sparke , l. , & de zeeuw , t. 1991 , apj , 373 , 369 bica , e. 1988 , a & a , 195 , 9 bushouse , h. a. 1986 , aj , 91 , 255 bowen , i. s. 1960 , apj , 132 , 1 bruzual , g. , & charlot , s. 2003 , mnras , 344 , 1000 castaneda , h. o. , vilchez , j. m. , & copetti , m. v. f. 1992 , a & a , 260 , 370 contini , t. , treyer , m. a. , sullivan , m. , & ellis , r. s. 2002 , mnras , 330 , 75 copetti , m. v. f.,mallmann , j. a. h. , schmidt , a. a. , & castaeda , h. o. 2000 , a & a , 357 , 621 dalcanton , j. j. 2007 , apj , 658 , 941 donzelli , c. j. , & pastoriza , m. g. 1997 , apjs , 111 , 181 dors , o. l. , jr .
, & copetti , m. v. f. 2005 , a&a , 437 , 837 dors , o. l. , jr . , & copetti , m. v. f. 2006 , a&a , 452 , 473 dutil , y. , & roy , j .-
r . 1999 ,
apj , 516 , 62 ellison , s. l. , patton , d. r. , simard , l. , & mcconnachie , a. w. 2008 , aj , 135 , 1877 ferland , g. j. 2002 , hazy , a brief introduction to cloudy 96.03 , univ .
kentucky , dept .
internal report ferreiro , d. l. , & pastoriza , m. g. 2004 , a & a , 428 , 837 ferreiro , d. l. , pastoriza , m. g. , & rickes , m. 2008 , a & a , 481 , 645 friedli , d. , benz , w. , & kennicutt , r. 1994 , apj , 430 , l105 garnett , d. r. , shields , g. a. , peimbert , m. , torres - peimbert , s. , skillman , e. d. , dufour , r. j. , terlevich , e. , & terlevich , r. j. 1999 , apj , 513 , 168 geller , m. j. , kenyon , s. j. , barton , e. j. , jarrett , t. h. , & kewley , l. j. 2006 , aj , 132 , 2243 gil de paz , a. , et al .
2007 , apj , 661 , 115 keenan , f. p. , hibbert , a. , ojha , p. c. , & conlon , e. s. 1993 , phys .
a. , 48 , 129 kennicutt , r. c. , jr . , & keel , w. c. 1984 , apj , 279 , l5 kennicutt , r. c. , jr . , roettiger , k. a. , keel , w. c. , van der hulst , j. m. , & hummel , e. 1987 , aj , 93 , 1011 kewley , l. j. , geller , m. j. , & barton , e. j. 2006 , aj , 131 , 2004 koeppen , j. , & arimoto , n. 1990 , a&a , 240 , 22 henry , r. b. c. , & worthey , g. 1999 , pasp , 111 , 919 hernquist , l. 1993 , apjs , 86 , 389 hernquist , l. , & mihos , j. c. 1995 , apj , 448 , 41 howarth , i. d. 1983 , mnras , 203 , 301 hummel , e. 1981 , a&a , 96 , 111 hummer , d. g. , & storey , p. j. 1987 , mnras , 224 , 801 iono , d. , yun , m. s. , & mihos , j. c. 2004 , apj , 616 , 199 joseph , r. d. , & wright , g. s. 1985 , mnras , 214 , 87 lamareille , f. , mouhcine , m. , contini , t. , lewis , i. , & maddox , s. 2004 , mnras , 350 , 396 lambas , d. g. , tissera , p. b. , alonso , m. s. , & coldwell , g. 2003 , mnras , 346 , 1189 larson , r. b. , & tinsley , b. m. 1978 , apj , 219 , 46 leitherer , c. , et al . 1999 ,
apjs , 123 , 3 kinkel , u. , & rosa , m. r. 1994 , a&a , 282 , l37 kronberger , t. , kapferer , w. , schindler , s. , bhm , a. , kutdemir , e. , & ziegler , b. l. 2006 , a & a , 458 , 69 mcgaugh , s. s. 1991 , apj , 380 , 140 mihos , j. c. , & hernquist , l. 1996 , apj , 464 , 641 nikolic , b. , cullen , h. , & alexander , p. 2004 , mnras , 355 , 874 noguchi , m. 1988 , a&a , 203 , 259 pagel , b. e. j. , edmunds , m. g. , blackwell , d. e. , chun , m. s. , & smith , g. 1979 , mnras , 189 , 95 pastoriza , m. g. , donzelli , c. j. , & bonatto , c. 1999 , a&a , 347 , 55 pilyugin , l. s. , contini , t. , & vlchez , j. m. 2004 , a&a , 423 , 427 pilyugin , l. s. , & thuan , t. x. 2005 , apj , 631 , 231 ramsbottom , c. a. , bell , k. l. , & stafford , r. p. 1996 , adndt , 63 , 57 roy , j .-
r . , & walsh , j. r. 1997 , mnras , 288 , 715 rubin , v. c. , waterman , a. h. , & kenney , j. d. p. 1999 , aj , 118 , 236 salpeter , e. e. 1955 , apj , 121 , 161 schmitt , h. r. , bica , e. , & pastoriza , m. g. 1996 , mnras , 278 , 965 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , apj , 500 , 525 sekiguchi , k. , & wolstencroft , r. d. 1992 , mnras , 255 , 581 springel , v. 2005 , mnras , 364 , 1105 struck , c. 1999 , phys . rep .
, 321 , 1 testor , g. 2001 , a&a , 372 , 667 trager , s. c. , worthey , g. , faber , s. m. , burstein , d. , & gonzalez , j. j. 1998 , apjs , 116 , 1 tremonti , c. a. , et al .
2004 , apj , 613 , 898 vale , t. b. , riffel , r. , pastoriza , m. g. , santiago , b. x. , rickes , m. g. , 2007 , in preparation vila - costas , m. b. , & edmunds , m. g. 1992 , mnras , 259 , 121 woods , d. f. , & geller , m. j. 2007 , aj , 134 , 527 worthey , g. , & ottaviani , d. l. 1997 , apjs , 111 , 377 zaritsky , d. , kennicutt , r. c. , jr . , & huchra , j. p. 1994 , apj , 420 , 87
the following supplementary material is available for this article : + * table 3 . *
radial velocities . +
* table 5 . * equivalent widths of the absorption lines . + * table 7 .
* dereddened line fluxes @xmath119 . +
this material is available as part of the online article from : http://www.blackwell-synergy.com/ ( this link will take you to the article abstract ) . | we present an observational study about the effects of the interactions in the kinematics , stellar population and abundances of the components of the galaxy pair am2306 - 721 .
rotation curves for the main and companion galaxies were obtained , showing a deprojected velocity amplitude of 175 km s@xmath0 and 185 km s@xmath0 , respectively .
the interaction between the main and companion galaxies was modeled using numerical n - body / hydrodynamical simulations , with the result indicating that the current stage of the merger would be about 250 myr after perigalacticum .
the spatial variation in the distribution of the stellar population components in both galaxies was analysed by fitting combinations of stellar population models of different age groups .
the central region of main galaxy is dominated by an old ( 5 - 10gyr ) population , while significant contributions from a young ( 200myr ) and intermediate ( 1gyr ) components are found in the disk , being enhanced in the direction of the tidal features .
the stellar population of the companion galaxy is overall much younger , being dominated by components with 1gyr or less , quite widely spread over the whole disk .
spatial profiles of the oxygen abundance were obtained from the a grid of photoionization models using the @xmath1 line ratio .
the disk of the main galaxy shows a clear radial gradient , while the companion galaxy presents an oxygen abundance relatively homogeneous across the disk .
the absence of an abundance gradient in the secondary galaxy is interpreted in terms of mixing by gas flows from the outer parts to the center of the galaxy due to the gravitational interaction with the more massive primary .
[ firstpage ] galaxies : general galaxies : stellar content galaxies : abundances galaxies : interactions galaxies : kinematics and dynamics galaxies : starburst |
the tensor decomposition problem arises in many applications ( see @xcite and references therein ) . because of many analogies with the matrix singular value decomposition ( svd )
, this multilinear generalization to high order tensors that we are going to consider , is often called higher - order singular vale decomposition ( hosvd ) " ( @xcite ) .
hosvd is a linear algebra method often used as a way to recover geometric or intrinsic informations , `` hidden '' in the tensor data . for a given tensor with a certain structure ,
this problem consists in finding the minimal decomposition into indecomposable tensors with the same structure .
the best known and studied case is the one of completely symmetric tensors ( see examples in @xcite , @xcite , @xcite ) , i.e. homogeneous polynomials .
the minimum number @xmath0 of indecomposable symmetric tensors @xmath1 s ( pure powers of linear forms @xmath2 s ) needed to write a given symmetric tensor @xmath3 of order @xmath4 ( that is a homogeneous polynomial @xmath5 of degree @xmath4 ) is called the _ rank
_ @xmath6 of @xmath3 ( the _ rank _ @xmath7 of @xmath5 ) : @xmath8 observe that when @xmath9 , i.e. when the tensor @xmath3 is a matrix ( i.e. when the homogeneous polynomial is a quadric ) , this coincides with the standard definition of rank of a matrix . in that case , a tensor decomposition of a symmetric matrix ( that can be obtained by svd computation ) of rank @xmath0 , will allow to write it as a linear combination of @xmath0 symmetric matrices of rank 1 . from now on , with an abuse of notation
, we will denote with `` @xmath5 '' both a symmetric tensor and its associate homogeneous polynomial . from a geometric point of view , saying that a symmetric tensor @xmath5 has rank @xmath0 , means that it is in the @xmath0-th secant of the veronese variety in the projective space of polynomials of degree @xmath4 .
the order @xmath10 of the smallest secant variety to the veronese variety containing a given @xmath5 is called the _ border rank _ of @xmath3 and may differs from the rank of @xmath5 ( see example [ ex : rk : monomial ] ) .
a first method to decompose a high order symmetric tensor is classically attributed to sylvester and it works for tensors @xmath11 with @xmath12 ( i.e. for binary forms ) . such a method ( see for a modern reference @xcite ) is based on the analysis of the kernel of so - called catalecticant matrices associated to the tensor .
this leads to the notion of _ catalecticant rank _
@xmath13 of a tensor @xmath5 , which is also called `` differential length '' in @xcite[definition 5.66 , p.198 ] . extending the apolarity approach of sylvester , an algorithm to compute the decomposition and the rank of a symmetric tensor @xmath5 in any dimension
was described in @xcite .
the main ingredient of this work is an algebraic characterization of the property of flat extension of a catalecticant matrix .
this extension property is not enough to characterize tensors with a given rank , since the underlying scheme associated to the catalecticant matrix extension should also be reduced . to get a better insight on this difference
, we introduce hereafter the notions of _ extension rank _ @xmath14 and _ border extension rank _
@xmath15 of @xmath5 , and analyze the main properties .
another approach leading to a different kind of algorithm is proposed in @xcite and it is developed for some cases .
the idea there , is to classify all the possible ranks of the polynomials belonging to certain secant varieties of veronese varieties in relation with the structure of the embedded non reduced zero - dimensional schemes whose projective span is contained in that secant variety . in @xcite , the authors clarify the structure of the embedded schemes whose span is contained in the secant varieties of the veronese varieties .
moreover they introduce an algebraic variety , namely the @xmath0-th cactus variety @xmath16 .
this lead us to the notion of what we will call the _ border scheme length _
@xmath17 of a polynomial @xmath5 .
we will show that this notion is related to the _ scheme length _ associated to @xmath5 defined in ( @xcite[definition 5.1 , p. 135 , definition 5.66 , p. 198 ] ) , we will call it the _ scheme length _ which is sometimes called the _ cactus rank _ of a homogeneous polynomial @xmath5 ( see @xcite for a first definition of it ) . another notion related to the scheme length and called the _ smoothable rank _
@xmath18 of a homogeneous polynomial @xmath5 is also used in @xcite[definition 5.66 , p. 198 ] or
@xcite . instead of considering all the schemes of length @xmath0 apolar to @xmath5 ,
one considers only the smoothable schemes , that are the schemes which are the limits of smooth schemes of @xmath0 simple points .
analogously we can define the _ border smoothable rank _
@xmath19 of a homogeneous polynomial @xmath5 , as the smallest @xmath0 such that @xmath5 belongs to the closure of the set of tensors of smoothable rank @xmath0 . in relation with the `` generalized additive decomposition '' of a homogeneous polynomial @xmath5
, there is the so called `` length of @xmath5 '' : it was introduced for binary forms in @xcite[definition 1.30 , p. 22 ] , and extended to any form in @xcite[definition 5.66 , p. 198
] . in this paper
we will describe a new generalization of the notion of _ generalized affine decomposition _ of a homogeneous polynomial @xmath5 ( see definition [ gendec ] ) and study the corresponding _ generalized rank _ @xmath20 .
again there is a notion of _ border generalized rank _
.08 in as in the classical tensor decomposition problem , the decompositions associated to these different notions of rank can be useful to analyze geometric information `` hidden '' in a high order tensor .
the purpose of this paper is to relate all these notions of rank .
this will give an algebraic geometric insight to a multilinear algebra concept as hosvd .
in corollary [ hierarchy ] we will show that the generalized rank , the scheme length and the flat extension rank coincide : @xmath22 and hence their respective border versions " : @xmath23 we can summarize the relations among the ranks in the following table : @xmath24 let @xmath25 , @xmath26 and @xmath27 be the sets of homogeneous polynomial of degree @xmath4 in a given number of variables of generalized rank , scheme length and extension rank respectively less than or equal to @xmath0 and let @xmath28 , @xmath29 and @xmath30 their zariski closures .
the main results of this paper is theorem [ prop : direct ] where we show that @xmath31 and hence ( corollary [ cor ] ) that @xmath32 .08 in the paper is organized as follows .
after the preliminary section [ preliminaries ] where we introduce some preliminary material on multilinear algebra and algebraic geometry needed for further developments , we will define , in section [ ranks ] , all the definitions of rank that we want to study and for each one of them we will give detailed examples . in sections [ gensection ]
we will prove our main results .
let @xmath33 $ ] be the graded polynomial ring in the variables @xmath34 over an algebraically closed field @xmath35 of characteristic 0 .
for @xmath36 , let @xmath37 be the the vector space spanned by the homogeneous polynomials of degree @xmath38 in @xmath39 .
we denote by @xmath40 $ ] the ring of polynomials in the variables @xmath41 and by @xmath42 the vector space of polynomials in @xmath43 of degree @xmath44 .
an ideal @xmath45 is _ homogeneous _ if it can be generated by homogeneous elements . for @xmath46 ,
we denote by @xmath47 the polynomial obtained by substituting @xmath48 by @xmath49 .
this defines a bijection between @xmath50 and @xmath51 , which depends on the system of coordinates chosen to represent the polynomials . for @xmath52 , we define @xmath53 and we call it the _ homogenization _ of @xmath5 .
a set @xmath54 of monomials of @xmath43 is _ connected to @xmath49 _ if it contains @xmath49 and if @xmath55 then there exists @xmath56 and @xmath57 such that @xmath58 . for a set @xmath54 of monomials in @xmath43 , @xmath59 .
we denote by @xmath60 the projective space of dimension @xmath61 . a point in @xmath62 which is the class of the non - zero element @xmath63 modulo the collinearity relation
is denoted by @xmath64 = ( k_{0 } : \cdots : k_{n})$ ] .
an ideal @xmath45 is homogeneous if it is generated by homogeneous polynomials . for a homogeneous ideal @xmath45 , the set of points @xmath64\in \pp^{n}$ ] such that @xmath65 is denoted @xmath66 .
we say that an ideal @xmath45 is _ zero - dimensional _ if @xmath66 is finite and not empty .
we say that @xmath67 is _ simple _ if the localization @xmath68 of @xmath69 at the maximal ideal @xmath70 associated to @xmath71 is of dimension @xmath49 ( cf .
an ideal @xmath72 of @xmath39 is _ saturated _ if @xmath73
. we will denote with @xmath74 the dedree @xmath4 part of an ideal @xmath72 .
the _ hilbert function _ associated to @xmath72 evaluated at @xmath36 is @xmath75 .
when @xmath72 is zero - dimensional , the hilbert function becomes equal to a constant @xmath76 for @xmath77 .
when moreover @xmath72 is saturated , this happens when @xmath78 ( see e.g. @xcite for more details ) . for a homogeneous ideal @xmath45 ,
let @xmath79 be the ideal of @xmath43 , generated by the elements @xmath80 for @xmath81 .
we recall that if @xmath82 for @xmath77 and if @xmath48 is a non - zero divisor in @xmath69 , then @xmath83 is a @xmath35-vector space of dimension @xmath0 .
conversely , if @xmath84 is an ideal of @xmath43 such that @xmath85 then the homogeneous ideal @xmath86 is saturated , @xmath48 is a non - zero divisor in @xmath69 and @xmath82 for @xmath78 .
if @xmath72 is a saturated ideal of @xmath39 and @xmath87 , then we have the natural isomorphism for @xmath36 : @xmath88 for a point @xmath89 , we define a corresponding element @xmath90 as @xmath91 .
the element @xmath92 is unique , up to a non - zero multiple : it corresponds to a unique element @xmath93 $ ] in @xmath94 . in the following
, we will use the same notation @xmath95 to denote either an element of @xmath96 or of @xmath97 .
the following product is sometimes called bombieri product " or sylvester product " .
[ apolarprod ] for all @xmath98 @xmath99 , we define the _ apolar product _ on @xmath50 as follows : @xmath100 where @xmath101 , @xmath102 for @xmath103 .
it can also be defined on @xmath51 in such a way that for all @xmath98 @xmath99 , @xmath104 ( just by replacing @xmath48 by @xmath49 in the previous formula ) . for any vector space @xmath105 , we denote by @xmath106 its _ dual space_. notice that the dual @xmath107 is an @xmath39-module : @xmath108 . for any homogeneous polynomial @xmath46
, we define the element @xmath109 as follows : @xmath110 similarly , @xmath111 is defined so that @xmath112 .
let @xmath72 be an ideal of @xmath39 .
the _ inverse system _ @xmath113 of @xmath72 is the @xmath39-submodule of elements of @xmath107 that vanish on @xmath72 , i.e. @xmath114 . for @xmath115 , we define @xmath116 as @xmath117 we check that if @xmath118 is a @xmath43-module , then @xmath119 is an ideal . when @xmath72 is a homogeneous ideal , an element in @xmath113 is a sum ( not necessarily finite ) of elements in @xmath120 . [ dimperp ]
the dimension of the degree @xmath4 part of the inverse system of an ideal @xmath45 is the hilbert function of @xmath69 in degree @xmath38 : @xmath121 we denote by @xmath122 the basis of @xmath123 that is dual to the standard monomial basis @xmath124 of @xmath125 , more precisely @xmath126 and @xmath127 if @xmath128 and @xmath129 otherwise .
an element in @xmath123 is represented by a homogeneous polynomial of degree @xmath4 in the _ dual variables _ @xmath130 .
it will also be called a _ dual polynomial_.
we remark that @xmath131 if @xmath132 and @xmath129 otherwise .
more generally , for any @xmath133 represented by a dual polynomial of degree @xmath4 , we have that @xmath134 is either @xmath129 or a dual polynomial of degree @xmath135 .
it is formally obtained by multiplying by @xmath136 and by keeping the terms with positive exponents .
this property explains the name of _ inverse system _ introduced by f.s .
macaulay @xcite .
the dual monomials are also called _ divided powers _ in some works , when a structure of ring is given to @xmath107 ( see e.g. @xcite[appendix a ] ) , but this structure is not really needed in the following .
it comes from the description of @xmath137 in terms of differentials : @xmath138 , @xmath139 where @xmath140 . for @xmath141
, we define the _ inverse system _ generated by @xmath118 as the @xmath39-module of @xmath107 generated by @xmath118 , that is the vector space spanned by the elements of the form @xmath142 for @xmath143 and @xmath144 .
the inverse system generated by @xmath145 is @xmath146 .
it is a vector space of dimension @xmath147 in @xmath148 $ ] . by extension
, the elements of @xmath107 can be represented by a formal power series in the variables @xmath149 . by restriction ,
the elements @xmath150 are represented by formal power series in the dual variables @xmath151 .
the elements of @xmath152 are represented by polynomials of degree @xmath153 in the variables @xmath151 .
the structure of @xmath43-module of @xmath150 shares the same properties as @xmath107 : @xmath154 acts as the `` inverse '' of @xmath155 .
we define the inverse system spanned by @xmath115 as the @xmath43-module of @xmath150 generated by @xmath118 . for a non - zero point @xmath156
, we define the evaluation @xmath157 at @xmath158 as @xmath159 in the following , we may drop the exponent @xmath4 to simplify notations when it is implicitly defined . to describe the dual of zero - dimensional ideals defining points with multiplicities , we need to consider differentials . for @xmath160 and @xmath161 , we defined @xmath162 we extend this definition by linearity , in order to define @xmath163 for any polynomial @xmath164 in the _ differential variables _ @xmath165 .
we check that the inverse system generated by @xmath166 is the vector space spanned by the elements of the form @xmath167 where @xmath168 is obtained from @xmath169 by possibly several derivations with respect to the differential variables @xmath165 .
it is a finite dimensional vector space .
this leads to the following result , which characterizes the dual of a zero - dimensional ( affine ) ideal ( see e.g. @xcite or @xcite[theorem 7.34 , p. 185 ] ) .
[ thm : dual : diff ] suppose that @xmath170 is such that @xmath171 .
then @xmath172 , there exist distinct points @xmath173 and differential polynomials @xmath174 in the variables @xmath175 such that @xmath176 as a consequence , we check that the inverse system generated by @xmath177 is the direct sum of the inverse systems @xmath178 generated by @xmath179 for @xmath180 .
the sum of the dimensions of these inverse systems is thus @xmath181 .
[ prop : mult : dual ] let @xmath182 , @xmath118 be the inverse system ( or @xmath43-module ) generated by @xmath177 and @xmath178 be the inverse system generated by @xmath183 for @xmath180 .
then @xmath184 where * @xmath185 is a primary ideal for the maximal ideal @xmath186 defining @xmath187 , * @xmath188 is the multiplicity of @xmath187 , * @xmath189 .
let us consider the ideal @xmath190 of @xmath191 $ ] .
it defines the points @xmath192 , @xmath193 , @xmath194 @xmath195 .
an element @xmath196 can be decomposed as @xmath197 where @xmath198 .
if @xmath199 , then the inverse system spanned by @xmath177 is @xmath200 [ lem : evdual ] suppose that @xmath72 is a saturated ideal defining @xmath0 simple points @xmath201 , \ldots , [ \zeta_{r } ] \in \pp^{n}$ ] .
then @xmath120 is spanned by @xmath202 for @xmath78 .
obviously @xmath203 .
moreover , as already observed in remark [ dimperp ] , we have that @xmath204 . therefore , for @xmath205 , @xmath206 and @xmath207 .
the main problem we are interested in , is the problem of _ decomposition of a symmetric tensor _ into a sum of minimal size of indecomposable terms which are the powers of a linear forms : [ def : decomp ] an element @xmath208 has a decomposition of size @xmath0 if there exist distinct non - zero elements @xmath209 such that @xmath210 this problem
is also called the _ generalized waring problem _ as it generalizes the problem of waring in arithmetic @xcite . in order to find a decomposition of @xmath211 as a sum of @xmath38-th powers of linear forms
, we will consider the polynomials which are apolar to @xmath5 and use the following result .
for all @xmath212 with @xmath213 , @xmath214 , for @xmath215 , it turns out that @xmath216 where @xmath217 . by an explicit computation
, we have @xmath218 thus @xmath219 thus if @xmath220 with @xmath221 and if @xmath222 is such that @xmath223 for @xmath224 , then for all @xmath225 we have @xmath226 this shows that the ideal of polynomials vanishing at the points @xmath227 is in the set of polynomials apolar to @xmath5 .
it leads us to the following definition ( see also @xcite where the same definition is given via an apolar product that differs from our definition [ apolarprod ] only because it is not defined as an inner product but as a product between @xmath125 and @xmath228 ) .
let @xmath211 .
we define the _ apolar ideal _ of @xmath5 as the homogeneous ideal of @xmath39 generated by @xmath229 and by the polynomials @xmath230 ( @xmath231 ) such that @xmath232 for all @xmath233 .
it is denoted @xmath234 . for @xmath235 with @xmath236 , we have @xmath237 and @xmath238 .
hereafter , we will need the following standard lemma .
[ lem : apolar : saturated ] for any ideal @xmath45 , @xmath239 if and only if @xmath240 . clearly , if @xmath240 then @xmath241 so that @xmath239 .
let us prove the reverse inclusion . by definition of the apolar ideal @xmath242 , we have @xmath243 .
we also have @xmath244 .
the hypothesis @xmath239 implies that @xmath245 .
we deduce that @xmath246 .
since @xmath247 , we have the inclusion @xmath248 . the tensor decomposition problem can be reformulated in terms of apolarity as follows via the well known apolarity lemma ( cf .
* lemma 1.15 ) ) .
[ prop : reform ] a symmetric tensor @xmath46 has a decomposition of size @xmath249 iff there exits an ideal @xmath45 such that * @xmath240 , * @xmath72 is saturated , zero dimensional , of degree @xmath250 , * @xmath72 is defining simple points .
suppose that @xmath5 has a decomposition of size @xmath250 : @xmath251 where @xmath252 and @xmath253 . then consider the homogeneous ideal @xmath72 of polynomials vanishing at the points @xmath254\in \pp^{n}$ ] , @xmath255 . by construction , for all @xmath256 , @xmath257 so that @xmath72 is a saturated ideal , defining @xmath258 @xmath259 simple points and with @xmath240 .
conversely , suppose that @xmath72 is an ideal of @xmath39 satisfying ( a ) , ( b ) , ( c ) .
let us denote by @xmath260 , \ldots , [ \mathbf{w}_{s}]$ ] the simple points of @xmath62 defined by @xmath72 and by @xmath261 corresponding elements in @xmath97 .
then by lemma [ lem : evdual ] , @xmath262 is spanned by @xmath263 . as @xmath240
, we have @xmath264 so that there exists @xmath265 such that @xmath266 this implies that @xmath267 and @xmath5 has a decomposition of size @xmath268 .
in this section we introduce all the different notions of rank of a homogeneous polynomial @xmath211 , that we will use all along the paper .
the following definition is nowadays a classical one , see e.g. @xcite and references therein .
[ def : secante2 ] let @xmath269 be the set of projective classes of homogeneous polynomials defined by @xmath270 \in \pp(s^{d } ) | \ \exists\ , \kb_1,\ldots,\kb_s \in s^{1 } \text { with } s \leq r \text { s.t . }
f = \kb_1^{d}+\cdots + \kb_s^{d } \}.\ ] ] for any @xmath211 , the minimal @xmath0 such that @xmath271\in \sigma_{r}^{0,{d}}$ ] is called the _ rank _ of @xmath5 and denoted @xmath7 .
[ ex : rk : monomial][eq : ex1 ] let us describe a decomposition of the monomial @xmath272 with @xmath236 of minimal size , which yields to its rank .
we consider the ideal @xmath273 for some @xmath274 .
it is defining @xmath275 simple points ( which @xmath276 coordinates are @xmath277 times the @xmath278-roots of unity ) .
let us consider the element @xmath177 of @xmath107 defined as follows : @xmath279 where @xmath187 is a primitive @xmath278-th root of unity for @xmath280 .
then for any monomial @xmath281 , we have @xmath282 where @xmath283 .
its minimal value on @xmath284 is @xmath285 .
the previous computation shows that @xmath286 suppose that @xmath287 .
then the ideal @xmath288 is included in @xmath289 . by proposition [
prop : reform ] , we deduce that @xmath5 has a decomposition of size @xmath290 .
as @xmath291 implies @xmath292 , we have @xmath293 which gives the decomposition of @xmath294 .
the corresponding decomposition of @xmath5 in terms of @xmath295-powers of linear forms is @xmath296 it can be proved that this decomposition has a minimal size ( see @xcite , @xcite ) , so that we have @xmath297 this example also shows that the decomposition is not unique , since @xmath277 is any non - zero constant . for more details on rank of monomials
see also @xcite , @xcite and @xcite ; the example above was also shown with different approach in @xcite and in ( * ? ? ?
* corollary 3.8 ) .
the zariski closure of @xmath298 , also known as the @xmath299 secant variety of the veronese variety of @xmath50 , is denoted @xmath300 .
the minimal @xmath0 such that @xmath271\in \sigma_{r}^{d}$ ] is called the _ border rank _ of @xmath5 and we denote it @xmath10 ( cf .
@xcite ) .
[ ex : border : rk : monomial][ex : ex2 ] consider again @xmath272 with @xmath236 .
suppose now that @xmath301 .
then the decomposition is of the form @xmath302 with possibly some terms in the sum which involves positive powers of @xmath277 .
this shows that @xmath303 . as @xmath304 and @xmath288
is defining simple points , the rank of @xmath305 is @xmath306 .
we deduce that the border rank of @xmath5 is less than @xmath307 .
in @xcite it is shown that if @xmath308 is equal to the sum of all the others @xmath309 s then such a bound is actually sharp .
+ consider e.g. @xmath310 ( for @xmath311 ) .
this is the first well known case where the rank and border rank are different : from example [ eq : ex1 ] we get that @xmath312 , while here we have just seen that @xmath313 ( see also @xcite , @xcite , @xcite ) .
let @xmath314 be the set of schemes of length @xmath0 which are the limit of smooth schemes of @xmath0 points , and let us consider the two following definitions according e.g. to @xcite and @xcite .
[ def : smoothable ] for any integers @xmath0 and @xmath4 , we define @xmath315 , as the set @xmath316 \in \mathbb{p } ( s^{{d } } ) | \ \exists \ , s \leq r , \exists\ , i \in \hilbredr , { \langle i^{{d } } , f \rangle}=0 \}.\ ] ] this leads to the following definition .
the smallest @xmath0 such that @xmath271 \in \smooth_{r}^{{d,0}}$ ] is called the _ smoothable rank _ of @xmath5 and it is denoted @xmath317 .
[ smooth : border ] in ( * ? ? ?
* lemma 5.17 ) it is shown that @xmath318 .
this proves that @xmath319 the following example is a personal communication from w. buczyska and j. buczyski ( @xcite ) .
it shows that strictly inequalities can occur .
[ jarek ] the following polynomial has border rank @xmath320 but smoothable rank @xmath321 : @xmath322 one can easily check that the following polynomial @xmath323 has rank @xmath324 for @xmath325 , and that @xmath326 . therefore @xmath327 . an explicit computation of @xmath328 yields to the following hilbert function for @xmath329 $ ] .
let us prove , by contradiction , that there is no saturated ideal @xmath330 of degree @xmath331 .
suppose on the contrary that @xmath72 is such an ideal .
then @xmath332 for all @xmath333 . as @xmath334 is an increasing function of @xmath335 with @xmath336
, we deduce that @xmath337 $ ] .
this shows that @xmath338 and that @xmath339 . as @xmath72 is saturated , @xmath340 since @xmath341 .
but an explicit computation of @xmath342 gives @xmath343 .
we obtain a contradiction , so that there is no saturated ideal of degree @xmath331 such that @xmath330 .
consequently , @xmath344 so that @xmath345 .
[ smooth = border ] if we indicate with @xmath346 the smallest @xmath0 such that @xmath271 \in \overline{\smooth_{r}^{{d,0}}}:=\smooth_{r}^{{d}}$ ] , then we can observe that @xmath347 .
obviously @xmath348 .
the other inclusion follows from remark [ smooth : border ] .
this shows that @xmath349 in the introduction we called @xmath346 the _ border smoothable rank _ of @xmath5 . the apolar ideal @xmath234 can also be defined via the kernel of the following operators .
let us recall the following standard definition .
[ def : catalecticant ] given a homogeneous polynomial @xmath208 and a positive integer @xmath350 such that @xmath351 , the _ catalecticant _ of order @xmath350 of @xmath5 , denoted by @xmath352 , is the application : @xmath353 its matrix in the monomial basis @xmath354 of @xmath355 and in the basis @xmath356 of @xmath357 is denoted @xmath358 . by construction , @xmath359 is the component @xmath360 of degree @xmath361 of the apolar ideal @xmath234 of @xmath5 . given two families of monomials @xmath362 and @xmath363 , we denote by @xmath364 the `` restriction '' of @xmath365 from the vector space spanned by @xmath54 to the dual of the vector space spanned by @xmath366 .
[ rem : symetrie.catalecticant ] by symmetry of the apolar product , we have @xmath367 via the identification @xmath368 . in terms of matrices
, we have @xmath369 orsalutalo da parte mia @xmath370 for all families of monomials @xmath371 and @xmath372 . [ cat ]
. the maximal rank of the operators @xmath374 , for @xmath375 , is called the _ catalecticant rank _ of @xmath5 and it is denoted @xmath376 .
this rank was already introduced in @xcite[definition 5.66 , p.198 ] where it was called `` the differential length of @xmath5 '' and denoted by @xmath377 .
[ def : variete.catalecticant ] given an integer @xmath378 and @xmath76 , we define the variety @xmath379 as : @xmath380 \in \mathbb{p } ( s^d ) | \ \rank(h_{f^{*}}^{i , d - i } ) = \rank(h_{f^{*}}^{d - i , i } ) \leq r \}.\ ] ] the set @xmath381 is the algebraic variety defined by the minors @xmath382 of the catalecticant matrices @xmath383 ( or @xmath384 ) .
these minors give not necessary reduced equations but they represents in @xmath385 the variety that is the union of linear spaces spanned by the images of the divisors ( hypersurfaces in @xmath386 ) of degree @xmath0 on the veronese @xmath387 ( see e.g. @xcite and @xcite ) . + if @xmath388 , such a variety is known as the `` subspace variety in @xmath385 '' @xmath389 .
for a generic @xmath390 , it can be geometrically obtained by intersecting @xmath385 with the @xmath0-th secant variety of the segre variety of @xmath391 ( for a better description of subspace varieties see @xcite ) . for a monomial @xmath392 with @xmath393 ,
the apolar ideal of @xmath5 is @xmath394 . by remark [ dimperp ] ,
the rank of @xmath395 is the dimension of @xmath396 that is the coefficient of @xmath397 in @xmath398 the maximum value of these coefficients which is the catalecticant rank is reached for the coefficients of the closest degree to @xmath399 ( it is proved in @xcite that the polynomial @xmath400 is symmetric unimodal , which means that its coefficients are increasing up to the median degree(s ) and then decreasing symmetrically ) . the exact value of the maximum is not known but asymptotic equivalents are known in some cases , see e.g. @xcite .
consider for instance the monomial @xmath401 .
the previous computation yields to the following hilbert series for the apolar ideal : @xmath402 this shows that @xmath403 , @xmath404 and thus that @xmath405 . according to example [
ex : ex2 ] , the border rank of @xmath401 is @xmath406 , which shows that the border rank of @xmath5 is strictly bigger that its catalecticant rank . in (
* theorems 1.2.3 and 4.2.7 ) , it is shown that @xmath407 has codimension @xmath324 in @xmath408 while the secant variety @xmath409 has codimension 6 .
therefore a generic element of @xmath407 has border rank strictly bigger than 5 .
[ gendec ] a generalized affine decomposition of size @xmath0 of @xmath46 is a decomposition of the form @xmath410 where @xmath411 and @xmath412 are differential polynomials , such that the dimension of the inverse systems spanned by @xmath413 is @xmath0 .
notice that the inverse system generated by @xmath414 is the direct sum of the inverse systems generated by @xmath415 for @xmath416 .
the inverse system generated by @xmath415 is the vector space spanned by the elements @xmath417 for all @xmath418 .
this decomposition generalizes the ( waring ) decomposition of definition [ def : decomp ] , since when @xmath419 are constant polynomials , we have the decomposition @xmath420 [ def : h ] given two integers @xmath0 and @xmath4 , we define @xmath421 by : @xmath422 \in pgl(n+1 ) } \ { [ f ] \in \mathbb{p } ( s^{d } ) \mid g\cdot f^ { * } \text { has a generalized affine decomposition of size } \le r \}.\ ] ] the smallest @xmath0 such that @xmath271 \in \gh_{r}^{{d,0}}$ ] is called the _ generalized rank _ of @xmath5 and it is denoted @xmath423 .
the polynomial @xmath424 defines an inverse system of dimension 4 obtained as @xmath425 , therefore @xmath426 .
moreover , we are in a case of a polynomial of border rank 4 and rank 7 ( as described in ( * ? ? ?
* theorem 44 ) ) . in this case @xmath427 .
[ expl : g0 ] for a monomial @xmath428 with @xmath429 , we have @xmath430 the inverse system spanned by @xmath431 is of dimension @xmath432 . assuming that @xmath433 , the previous decomposition is a generalized decomposition of minimal size ( according to corollary [ hierarchy ] and example [ ex : schematic : rk : monomial ] ) .
therefore we have @xmath434 notice that @xmath271 \in \gh_{r}^{{d,0}}$ ] iff there exists a change of coordinates such that in the new set of coordinates @xmath435 has a generalized affine decomposition of size @xmath436 .
this notion of generalized affine decomposition and of generalized rank is related to the generalized additive decomposition introduced in ( * ? ? ?
* definition 1.30 , p. 22 ) for binary forms , called `` the length of @xmath5 '' and denoted @xmath437 .
however , the extension to forms in more variables proposed in ( * ? ? ?
* definition 5.66 , p. 198 ) does not correspond to the one we propose , in fact it corresponds to the border rank . for
binary forms , the border rank and the generalized rank coincide as we will see in the sequel .
given two integers @xmath0 and @xmath4 , we define @xmath438 to be the zariski closure of @xmath439 defined above .
the smallest @xmath0 such that @xmath271 \in \gh_{r}^{{d}}$ ] is called the _ border generalized rank _ of @xmath5 and it is denoted @xmath440 .
we describe here a new notion of rank based the property of extension of bounded rank of the catalecticant matrices . for any integers @xmath0 and @xmath4 ,
we define @xmath441 , as the set @xmath442
\in \mathbb{p } ( s^{{d } } ) | \ \exists \ , \ub \in s^{1}\setminus\{0\ } , \exists\ , [ \tilde{f } ] \in \gamma^{m , m'}_{r}\ \mathrm{\ with\ } \\ & & \ \ m = \max\{r , \lceil { d\over 2}\rceil\ } , \ m ' = \max\{r-1 , \lfloor { d\over 2}\rfloor\ } \ \mathrm{s.t.}\ \ub^{m+m'-d}\cdot \tilde{f}^ { * } = f^{*}\}.\end{aligned}\ ] ] by definition , @xmath443 .
moreover , if @xmath444 then @xmath445 since @xmath446 , @xmath447 and @xmath448 . the smallest @xmath0 such that @xmath271 \in \eh_{r}^{{d,0}}$ ] is called the _ flat extension rank _ of @xmath5 and it is denoted @xmath449 . a @xmath450 \in \gamma_{r}^{m , m'}$ ] such that @xmath451 with @xmath452 and @xmath453 is called a _ flat extension _ of @xmath454 of rank @xmath0 . for a monomial @xmath428 with @xmath429 , the element @xmath455 defines , by restriction , a hankel operator @xmath456 from @xmath457 to @xmath458 where @xmath459 .
we check that the image of @xmath456 is the vector space of @xmath458 spanned by @xmath460 for @xmath461 .
thus @xmath456 is of rank @xmath0 .
according to example [ expl : g0 ] and theorem [ thm : cactus.catalecticant ] , @xmath462 is a flat extension of @xmath5 of minimal rank when @xmath463 .
we deduce that @xmath464 notice that @xmath271 \in \eh_{r}^{{d,0}}$ ] iff there exists a change of coordinates such that after this change of coordinates we have @xmath465 so that @xmath466 is such that * @xmath467 and * @xmath468 . in other words ,
@xmath271 \in \eh_{r}^{{d,0}}$ ] iff after a change of coordinates , @xmath469 can be extended to a linear form @xmath466 with @xmath470 .
we will see hereafter that we can choose a generic change of coordinates . a simple way to characterize a flat extension of a given rank
is given by the following result .
[ thm : decomp.generalise.extension ] let @xmath471 be sets of @xmath43 , @xmath472 and @xmath473 be two monomial sets of size @xmath0 connected to @xmath49 such that @xmath474 contains @xmath475 . if @xmath476 is such that @xmath477 , then there exists an extension @xmath478 of @xmath177 such that @xmath479 .
moreover , we have @xmath480 . for any integers @xmath0 and @xmath4 , we define @xmath481 , as the zariski closure of set @xmath482 defined above and the smallest @xmath0 such that @xmath271 \in \eh_{r}^{{d}}$ ] is called the _ border flat extension rank _ of @xmath5 and it is denoted @xmath483 .
we recall that @xmath484 is the set of 0-dimensional schemes @xmath485 of @xmath258 points ( counted with multiplicity ) .
it can be identified with the set of homogeneous saturated ideals @xmath45 such that the algebra @xmath69 has a constant hilbert polynomial equal to @xmath258 .
[ def : cactus ] for any integers @xmath0 and @xmath4 , we define @xmath486 , as the set @xmath487 \in \mathbb{p } ( s^{{d } } ) | \ \exists \ , s \leq r , \exists\ , i \in \hilbs , { \langle i^{{d } } , f \rangle}=0 \}.\ ] ] in @xcite , the _
@xmath0-cactus variety _
@xmath16 is defined as the closure of @xmath488 .
the smallest @xmath0 such that @xmath271 \in \kak_{r}^{{d,0}}$ ] is called the _ scheme length _ ( or _ cactus rank _ in @xcite ) of @xmath5 and it is denoted @xmath489 .
the smallest @xmath0 such that @xmath271 \in \kak_{r}^{{d}}$ ] is called the _ border scheme length _ of @xmath5 and it is denoted @xmath17 .
we have used the same definition of scheme length " of @xmath5 used in @xcite[definition 5.1 , p. 135
, definition 5.66 , p. 198
] , where it is denoted @xmath490 .
[ ex : schematic : rk : monomial ] for a monomial @xmath428 with @xmath429 , the ideal @xmath491 is an ideal of length @xmath492 , which is apolar to @xmath5 . assuming that @xmath463 , this length @xmath493 is minimal as proved in ( * ? ? ?
2 ) , using bezout theorem .
thus , the scheme length of @xmath5 is : @xmath494 this is an example where the border rank and scheme length coincide but they differ from the rank ( see example [ ex : border : rk : monomial ] ) . in the next example , we have a case where the scheme length is strictly smaller that the border rank . in the case of cubic polynomials ,
the scheme length of a generic form is smaller than its border rank for forms in 9 variables .
the border rank of a generic cubic form in 9 variables is in fact 19 ( this is alexander and hirschowitz theorem @xcite ) , while the scheme length is smaller or equal than 18 ( see @xcite ) .
the objective of this section is to relate the scheme length , generalized rank and flat extension rank . by a change of coordinates
, we can assume that @xmath502 .
then the linear form : @xmath503 coincides with @xmath504 on @xmath505 . by proposition [ prop : mult : dual ] , as the dimension of the inverse system generated by @xmath177 is @xmath249 , @xmath506 is a zero - dimensional ideal of multiplicity @xmath507 .
we denote by @xmath508 the homogenization of @xmath509 with respect to @xmath510 . then
, @xmath511 as @xmath512 on @xmath505 , we have : @xmath513 which proves the inclusion .
let us prove that for all @xmath516 such that @xmath239 with @xmath517 and @xmath249 , we have : @xmath518 for all @xmath519 . by lemma [ lem : apolar : saturated ] , we have @xmath520 so that @xmath521 for @xmath519 .
as the hilbert function of a saturated ideal @xmath522 is increasing until degree @xmath258 and then it is constantly equal to @xmath258 , we have @xmath523 by the above inclusion , this implies that @xmath524 as @xmath525 , we deduce that @xmath526 consequently as @xmath507 , @xmath527 and @xmath528 [ cor : jerome ] let @xmath78 and @xmath532 such that @xmath533 is of dimension @xmath0 .
then for a generic change of coordinates @xmath534 , @xmath535 has a monomial basis of the form @xmath536 with @xmath537 .
moreover , @xmath538 is connected to @xmath49 .
let @xmath539 be the lexicographic ordering such that @xmath540 .
by @xcite[theorem 15.20 , p. 351
] , after a generic change of coordinates , the initial @xmath541 of the ideal @xmath542 for @xmath539 is borel fixed .
that is , if @xmath543 then @xmath544 for @xmath545 . to prove that there exists a subset @xmath54 of monomials of degree @xmath135 such that @xmath546 is a basis of @xmath547 ,
we show that @xmath548 .
let @xmath549 , @xmath550 $ ] and @xmath551 .
then we have the exact sequence @xmath552 where @xmath553 is the multiplication by @xmath48 .
let us denote by @xmath554 and @xmath555 for @xmath556 .
suppose that @xmath557 , then @xmath558 . as @xmath78 and @xmath0 is the gotzmann regularity of @xmath559 , by (
* ( 2.10 ) , p. 66 ) we have @xmath560 .
as @xmath541 is borel fix , i.e. @xmath561 implies @xmath562 for @xmath563 , we have @xmath564 , so that @xmath565 .
this implies that @xmath566 , which contradicts the hypothesis on @xmath105 .
thus @xmath548 .
as @xmath567 is borel fix and different from @xmath50 , its complementary @xmath366 contains @xmath570 . similarly we check that if @xmath571 with @xmath572 and @xmath573 then @xmath574 .
this shows that @xmath575 is connected to @xmath49 .
let @xmath577 and @xmath580 .
we first prove the following inclusion : @xmath581 let us fix @xmath271 \in \gamma_{r}^{i , d - i}$ ] for an integer @xmath582 .
let us denote @xmath583 and @xmath584 and @xmath585 the rank of @xmath586 .
we recall that : @xmath587 the quotients @xmath588 and @xmath589 are thus of dimension @xmath361 . as @xmath590 and @xmath591 , by lemma [ cor : jerome ] and a generic change of coordinates we may assume that there exists a family @xmath54 ( resp .
@xmath366 ) of @xmath361 monomials of @xmath592 ( resp .
@xmath593 ) such that @xmath594 ( resp .
@xmath595 ) is a basis of @xmath588 ( resp .
@xmath596 ) and that @xmath597 are connected to @xmath49 . notice then that @xmath598 is an invertible matrix of size @xmath599 . as the monomials of @xmath538 are in @xmath600 ( resp .
@xmath601 ) , the monomials of @xmath602 ( resp .
@xmath603 ) are in the set @xmath604 ( resp .
@xmath605 ) of monomials of degree @xmath606 ( resp .
@xmath607 ) and @xmath608 .
moreover , we have @xmath609 by theorem [ thm : decomp.generalise.extension ] , there exists a linear form @xmath610 which extends @xmath611 such that @xmath612 where @xmath613 . by theorem [ thm : dual : diff ] , as @xmath614 , there exists @xmath615 , differential polynomials @xmath616 $ ] such that @xmath617 as the inverse system spanned by @xmath177 is included in @xmath618 which is of dimension @xmath258 and as @xmath177 coincides with @xmath611 on @xmath51 , @xmath5 has a generalized affine decomposition of size @xmath268 and @xmath271 \in \gh_{r}^{{d}}$ ] .
this proves that @xmath619 using lemmas [ lem : gink ] and [ lem : kingamma ] , we have @xmath620 by definition , @xmath621 when @xmath622 .
as @xmath623 is closed , the previous inclusions show that @xmath624 [ lem : kinh ] given @xmath0 , @xmath4 @xmath629 , @xmath630 , @xmath211 and @xmath631 such that @xmath632 is a non - zero divisor for @xmath72 , there exists a flat extension @xmath633 of @xmath5 such that @xmath452 , and we have @xmath634 .
we denote by @xmath640 the dehomogenization of @xmath641 ( setting @xmath642 ) . as @xmath637 and @xmath87 ,
the quotient algebra @xmath83 is a @xmath643-vector space of dimension @xmath0 . by the following natural isomorphism : @xmath644 a linear form @xmath645 corresponds to a linear form @xmath646 . as @xmath647
the linear form @xmath646 can be extended ( not necessary in a unique way ) to a linear form @xmath648 .
as the ideal @xmath649 contains @xmath79 and @xmath650 , we deduce that @xmath651 is of rank @xmath436 . by restriction
, @xmath652 is of rank @xmath436 , where @xmath653 .
the restriction of @xmath654 to @xmath51 is @xmath504 . for any @xmath271 \in \kak_{r}^{d,0}$ ] , there exists @xmath631 such that @xmath636 . as @xmath72 is a saturated ideal , it is always possible to find @xmath655 such that @xmath656 is non - zero divisor for @xmath72 .
by the previous construction , we can find a flat extension @xmath450 \in \gamma^{m+m'}_{r}$ ] of @xmath271 $ ] , which this shows that @xmath271\in \eh_{r}^{{d,0}}$ ] and that @xmath657 .
* there exists a zero - dimensional saturated ideal @xmath72 defining @xmath436 points counted with multiplicity such that @xmath658 ; * @xmath5 has a generalized decomposition of size @xmath436 ; * @xmath5 has a flat extension of size @xmath436 .
let us prove that @xmath663 . for any @xmath271\in \eh_{r}^{{d,0}}$
] , there exists @xmath450\in \gamma^{m , m}_{r}$ ] and @xmath630 such that @xmath664 . by theorem [ thm : cactus.catalecticant ] , @xmath450\in \gh_{r}^{2m,0}$ ] .
thus after some change of coordinates , @xmath462 has an affine generalized decomposition of the form : @xmath665 as @xmath666 , we have @xmath667 where @xmath668 is obtained from @xmath669 by derivation .
as @xmath668 is obtained from @xmath669 by derivation , the inverse system spanned by @xmath670 is included in the inverse system spanned by @xmath671 .
thus @xmath5 has an affine decomposition @xmath672 of size @xmath250 and @xmath271\in \gh_{r}^{{d,0}}$ ] .
form corollary [ kim : rank ] we get that @xmath529 . by definitions of @xmath678 , @xmath679 and @xmath680
we obviously have that @xmath681 , @xmath682 , @xmath683 .
finally theorem [ prop : direct ] and corollary [ cor ] end the proof . at the beginning of the proof of theorem [ thm : cactus.catalecticant
] , we showed that if @xmath577 and @xmath580 , then @xmath688 , @xmath689 and @xmath690 observe that in example [ jarek ] the condition @xmath577 is not satisfied .
since , by corollary [ hierarchy ] , we have that @xmath691 , then example [ jarek ] gives also an example of a polynomial @xmath5 such that @xmath692 .
we thank a. bostan for the precise and detailed information he provided on combinatoric results related to the estimation of the catalecticant rank .
we also thank j. buczyski for many useful discussions and suggestions . | we introduce various notions of rank for a high order symmetric tensor , namely : rank , border rank , catalecticant rank , generalized rank , scheme length , border scheme length , extension rank and smoothable rank .
we analyze the stratification induced by these ranks .
the mutual relations between these stratifications , allow us to describe the hierarchy among all the ranks .
we show that strict inequalities are possible between rank , border rank , extension rank and catalecticant rank .
moreover we show that scheme length , generalized rank and extension rank coincide . |
waves and oscillations are a common phenomeno in solar flares .
they are believed to reveal themselves as quasi - periodic pulsations in spatially unresolved observations at different wavelengths ( see , _ e.g. _ , the review of * ? ? ?
* ) and have been directly detected using the euv observations with high spatial resolution @xcite .
in particular , the observed periodicities of the quasi - periodic pulsations in the radio emission , hard x - rays , and @xmath0-rays ( from a fraction of a second to several minutes ) agree with the expected magnetohydrodynamic ( mhd ) timescales which strongly suggests a close relationship between the pulsations and the mhd waves and oscillations @xcite .
it should be noted , however , that the solar quasi - periodic pulsations are usually superimposed on a rapidly varying background , so that recovering their parameters is not easy ; some previously reported conclusions may be unreliable due to the methodology used @xcite .
observations of the waves and oscillations in the flares can be used to diagnose the quasi - periodic magnetic reconnection processes that are believed to be the most probable source of these oscillations ( _ e.g. _ , * ? ? ?
the mhd waves and oscillations can also be used to diagnose the coronal plasma and magnetic field parameters ( coronal seismology ; see , _ e.g. _ , * ? ? ?
moreover , the waves and oscillations are believed to play a key role in the energy transfer both in the quiet corona ( _ e.g. _ , * ? ? ?
* ) and in the flares @xcite ; they can modulate and/or trigger the magnetic reconnection processes @xcite and thus affect greatly the behaviour of unstable active regions and development of the solar flares .
one of the potential observable manifestations of mhd waves is quasi - periodic pulsations of radio emission ( _ e.g. _ , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , because these waves can modulate both the coherent and incoherent radio emission mechanisms .
interpretation of the observed quasi - periodic pulsations requires comparing the observations with simulations in order to determine the origin of the pulsations ( because they can be caused not only by the mhd waves but also , _
e.g. _ , by quasi - periodic injection of energetic particles ) ; if the oscillations are due to waves , it allows us to identify the wave mode and other characteristics of the mhd waves . in turn ,
identification of the wave mode is of high importance for the development of the coronal seismology . as a rule ,
the basic characteristics of the pulsations ( such as the period ) have no unique interpretation and therefore some more subtle features ( such as the relations between the pulsations observed in different spectral channels ) have to be analyzed @xcite .
perhaps the best tool to study the physical parameters and processes in solar flares is their microwave emission , because it is produced mainly due to the incoherent gyrosynchrotron radiation mechanism which is ( a ) relatively simple and well studied @xcite and ( b ) highly sensitive to the magnetic field and plasma parameters .
there are a lot of papers dedicated to simulating the gyrosynchrotron emission in solar flares .
in particular , two - dimensional ( 2d ) models ( _ e.g. _ , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and three - dimensional ( 3d ) models ( _ e.g. _ , * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have been employed to account for the structure of the emitting region .
some simulations involve the magnetic field structure obtained by extrapolating the observed photospheric magnetograms ( _ e.g. _ , * ? ? ?
* ; * ? ? ?
* ) , take into account anisotropy of the emitting electrons ( _ e.g. _ , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and/or evolution of the electron distributions during flares ( _ e.g. _ , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , possible modulation of the emission by mhd waves has been studied only in a few works and by using relatively simple models @xcite .
recently , @xcite have simulated the gyrosynchrotron radio emission from an oscillating magnetic cylinder ( with standing sausage waves ) .
it was found that even for the simple model considered ( a straight cylinder ) , calculating the emission requires 3d simulations that also take into account the spatial structure of the mhd waves , because the results can differ from simple estimates for a spatially homogeneous source ( _ cf .
* ; * ? ? ?
at the same time , another complication arises from the fact that the solar flaring loops are ( in contrast to the model of @xcite ) evidently curved .
this work is dedicated to simulating the radio emission from such curved loops .
simulation of mhd waves in flaring loops is a nontrivial task .
oscillation modes of a straight overdense cylinder of infinite length are well studied ( * ? ? ?
* ; * ? ? ?
* _ etc . _ ) , and the numerical simulations can be greatly simplified by using the analytical results
. however , these results are not applicable to curved loops @xcite .
variation of the plasma and magnetic field parameters as well as of the loop radius with height presents even more difficulties @xcite .
therefore in this work we use a simplified model in which an oscillating cylinder is artificially bent to form a semi - circular loop without actually using the wave relations in a semi - torus ; the magnetic field , plasma density and temperature , and the loop radius ( in the equilibrium state , _
i.e. _ , without oscillations ) are assumed to be constant along the loop .
the main aim of this model is to study the effect of varying the viewing angle ( _ i.e. _ , the angle between the local magnetic field and the line of sight ) along the loop .
although such a model can not correctly describe the entire loop , it can reproduce the effects of the mhd oscillations in local parts of the loop , where the effects of the loop curvature and inhomogeneity on the mhd wave parameters are negligible .
we consider the incoherent gyrosynchrotron radiation of energetic electrons which makes the main contribution to the flare emission in the microwave range .
the simulation model is described in section [ model ] .
the simulation results are presented and discussed in section [ results ] .
the conclusions are drawn in section [ conclusion ] .
we use the model described earlier by @xcite , @xcite , @xcite , _ etc . _ in this model , there is an overdense cylinder with the magnetic field directed along its axis .
the plasma density and temperature inside the cylinder are higher than outside , while the magnetic field strength inside the cylinder is lower than outside , to maintain the pressure balance .
we consider the azimuthally symmetric case when all parameters depend only on the distance along the cylinder axis @xmath1 and the distance from this axis @xmath2 . in an equilibrium case , the magnetic field , plasma density , and temperature are assumed to be constant inside and outside the cylinder ; they experience a sharp jump at the cylinder boundary ( at @xmath3 ) .
the formulae describing the mhd disturbances are given , e.g. , in the works of @xcite , @xcite , @xcite , and @xcite ; see especially the paper of @xcite where the solutions of those equations are presented for the same conditions as used in this work .
therefore we describe the properties of the mhd waves only briefly .
we consider the sausage mode the compressible azimuthally symmetric ( @xmath4 ) oscillations that affect both the magnetic field and plasma density and temperature .
we consider both the standing and propagating waves which are described by the relations @xmath5 respectively , where @xmath6 is the wave amplitude ( _ e.g. _ , in velocity ) , @xmath7 and @xmath8 are the wave frequency and wavenumber , respectively , and @xmath9 is the initial phase . ) is considered ; one wavelength is shown .
profiles of the plasma density and temperature are similar to those of the magnetic field.,title="fig : " ] ) is considered ; one wavelength is shown .
profiles of the plasma density and temperature are similar to those of the magnetic field.,title="fig : " ] figure [ figmhd ] presents an example of how the parameters vary with the coordinates in a standing wave ( the oscillation phase is chosen to provide the maximum deviation from equilibrium ) . in general , the axial magnetic field @xmath10 , plasma density @xmath11 , and temperature @xmath12 vary in phase and their variations are maximal at the cylinder axis ( we are interested only in the parameters inside the cylinder , because , as shown below , this is the region where the radio emission is assumed to be generated ) .
variations of the above parameters decrease with increasing @xmath2 and become zero at some radial distance @xmath13 nearby the cylinder boundary ( @xmath14 for the parameters considered here )
. then the amplitude of the variations increases slightly , but they are now in anti - phase with the variations of the same parameters in the inner part of the cylinder .
variations of the radial magnetic field @xmath15 are shifted by @xmath16 period with respect to other parameters ; they are zero at the cylinder axis and reach their maximum at @xmath17 ( without the phase reversal at that radial distance ) . since we consider low - amplitude oscillations , we have @xmath18 ( we remind that @xmath10 includes here both the equilibrium magnetic field @xmath19 and the perturbed field component @xmath20 , _
i.e. _ , @xmath21 ) .
the radial component @xmath22 makes almost no contribution to the resulting magnetic field strength ; however , it affects slightly the magnetic field direction causing it to deviate from the cylinder axis .
variations of the cylinder radius due to the mhd oscillations are very small and do not affect the radio emission .
the propagating sausage wave looks like the structure shown in figure [ figmhd ] moving along the @xmath1 axis .
all other properties of this wave ( including the phase relations between different parameters ) are the same as for the standing wave . lcc model & low - density & high - density + + @xmath23 [ km ] & 1000 & 1000 + @xmath24 [ g ] & 160 & 50 + @xmath25 [ g ] & 161 & 56 + @xmath26 [ @xmath27 & @xmath28 & @xmath29 + @xmath30 [ @xmath27 & @xmath31 & @xmath32 + @xmath33 [ k ] & @xmath34 & @xmath34 + @xmath35 [ k ] & @xmath36 & @xmath36 + @xmath37 [ @xmath27 & @xmath38 & @xmath39 + + @xmath40 [ km ] & 2825 & 2825 + @xmath41 [ s ] & 0.35 & 1.5 + @xmath42 [ g ] & @xmath43 & @xmath44 + @xmath45 & @xmath46 & @xmath47 + @xmath48 [ @xmath27 & @xmath49 & @xmath50 + @xmath51 [ k ] & @xmath52 & @xmath53 + @xmath54 [ @xmath27 & @xmath55 & @xmath56 + we consider two sets of model parameters , the same as have been used in the work of @xcite ;
the parameters are listed in table [ tabmhd ] .
the parameters were chosen to provide the same wavelength @xmath40 for the given cylinder radius @xmath23 . in the calculations below ,
we use mainly the low - density model because those conditions are more favorable for generation of the gyrosynchrotron radio emission . in the high - density model ,
the emission intensity is usually lower ; at low frequencies , the emission becomes strongly affected by the razin effect ( see section [ razin ] ) . as stated before ,
the aim of this work is to study the effect of the loop curvature on the radio emission . for this , we have created model semi - circular ( semi - toroidal ) loops by bending the above - mentioned overdense cylinder ( see figure [ figloop ] ) . in this case , the longitudinal coordinate within the cylinder @xmath1 is transformed into the distance along the loop axis , and the radial coordinate @xmath2 into the distance from the loop axis ( the explicit formulae are given in the appendix ) ; this allows us to obtain the plasma and magnetic field parameters at any given point .
since the lengths of the solar flaring loops are usually much larger than their radii , we have created the models by stacking several mhd simulation blocks ( each corresponding to one wavelength , like in figure [ figmhd ] ) along the loop . as the `` basic '' model
, we have used the loop containing three wavelengths ; such loop has the length of 8475 km and the height ( _ i.e. _ , the curvature radius ) of about 2700 km .
the loop orientation is described in general by three euler angles . however , since rotation around the line of sight does not provide us with any new information ( it results simply in the same rotation of the simulated radio images ) , it is sufficient to consider the variation of only two angles @xcite . in this work
we assume that the loop is vertical and located at the solar equator .
the loop orientation is described by two angles : the angle @xmath57 between the loop plane and the equatorial plane and the longitude @xmath40 .
a possible loop inclination is not considered for the same reason as described above : for a vertical loop , the variation of the angles @xmath57 and @xmath40 is sufficient to obtain any possible loop orientation with respect to the observer ( with an accuracy up to rotation around the line of sight ) ; therefore our results are applicable to inclined loops as well
. however , we note that the loop locations at the solar disk which are discussed below should be treated as approximate ( illustrative ) , because they correspond to the zero inclination only . ) along the loop axis ( bottom row ) .
the visible coordinate along the loop @xmath58 is measured from the upper ( as shown in the 3d views ) footpoint . ] ) along the loop axis ( bottom row ) .
the visible coordinate along the loop @xmath58 is measured from the upper ( as shown in the 3d views ) footpoint . ]
it is not possible to present here the results for all possible orientations ; therefore we have focused on three illustrative cases ( see figure [ figloop ] and table [ tabloops ] ) : \a ) the loop is located near the disk center . in this case , the viewing angle @xmath59 ( relative to the magnetic field ) varies in a wide range along the loop , being close to zero ( or @xmath60 ) in the footpoints and being equal to @xmath61 at some point near the loop top ( see figure [ figloop ] ) .
the coordinate @xmath58 in figure [ figloop ] is the visible distance along the loop axis , _ i.e. _ , it takes into account the projection effects .
\b ) the loop is located near the solar limb and its plane is almost parallel to the equatorial plane ( `` limb a '' orientation ) . in this case , in contrast to the previous one , the viewing angle is close to @xmath61 in the loop footpoints and becomes very small near the loop top .
\c ) the loop is located near the solar limb and its plane is almost perpendicular to the line of sight ( `` limb b '' orientation ) . in this case
the viewing angle is always close to @xmath61 and weakly varies along the loop ( although it is not constant , which makes this model different from those studied in the work of @xcite ) .
.orientations of the magnetic loop used in the simulations : the angle relative to the equatorial plane ( @xmath57 ) and the longitude ( @xmath40 ) .
[ cols="^,^,^,^",options="header " , ] we assume that the gyrosynchrotron radio emission is produced by the energetic electrons with the power - law distribution over energy , @xmath62 ; in the calculations , the energy range is taken to be @xmath63 which is typical for the solar flares .
the electrons with lower energies are also present in flares , but their contribution to the gyrosynchrotron emission is minor ( mostly in the form of self - absorption at low frequencies ; see , _ e.g. _ , @xcite ) and usually can be ignored .
the pitch - angle distribution of the electrons is assumed to be isotropic ( _ i.e. _ , we assume that it is not affected significantly by the mhd waves ) . @xcite
have considered the case when the concentration of the energetic electrons @xmath64 inside the cylinder is proportional to the thermal plasma density , @xmath65 .
alternatively , one may consider the model in which the energetic electrons are `` frozen '' into the magnetic field , so that their concentration is proportional to the field strength , @xmath66 .
however , since in the sausage waves the dominant longitudinal component of the magnetic field @xmath10 and the plasma density @xmath11 oscillate in phase , both approaches provide almost identical results .
therefore in the calculations to follow we assume that the concentration of the energetic electrons in the loop is always proportional to the thermal plasma density and @xmath67 ; in particular , in the equilibrium state we have @xmath68 and @xmath39 @xmath69 for the low- and high - density models , respectively .
we assume that there are no energetic electrons outside the loop ( @xmath70 ) .
the gyrosynchrotron radio emission is calculated using the fast gyrosynchrotron codes @xcite ; these codes include the contribution of the thermal free - free emission as well .
we neglect the possible harmonic structure of the gyrosynchrotron emission at low frequencies by using the `` continuous '' regime of the fast gyrosynchrotron codes , because in real observations this structure should be usually smoothed by the source inhomogeneity @xcite ; using the `` continuous '' approximation also speeds up the calculations appreciably .
the codes calculate the emission characteristics by numerical integration of the radiative transfer equations for the ordinary and extraordinary electromagnetic modes while taking into account the possible mode - coupling effects , _
i.e. _ , interaction between the modes in the regions of transverse magnetic field @xcite .
this approach is sufficient to reproduce accurately the emission intensity in both the quasi - longitudinal and quasi - transverse ( with respect to the magnetic field ) propagation regimes as well as the circular polarization in the quasi - longitudinal regime @xcite , although simulating the polarization in the regions with @xmath71 may require considering the full set of stokes parameters ( see the upcoming paper of @xcite ) .
only the emission and absorption processes inside the curved magnetic loop are considered , _
i.e. _ , the emission outside the loop is assumed to propagate like in vacuum . and
6 ) and the `` limb b '' orientation are considered . ]
firstly , we present a brief overview of the emission produced by the loop in the equilibrium state ( _ i.e. _ , without mhd oscillations ) .
this emission is very similar to the emission from an oscillating loop averaged over the wave period .
figure [ figspectrum ] demonstrates the total ( spatially integrated ) intensity spectra for the loop located at the limb ( `` limb b '' orientation ) . for other orientations ,
the intensity is slightly lower , but the spectra shapes are qualitatively the same .
the concentration of the energetic electrons was chosen to provide that the spectral maximum ( for the low - density model and the spectral index of the energetic electrons of @xmath72 ) is located at about 10 ghz a typical value for the continuum microwave emission of solar flares .
the bottom row in figure [ figloop ] demonstrates the simulated distributions of the gyrosynchrotron radio intensity in the equilibrium state along the loop ( for the low - density model ) .
the intensity distributions depend on the emission frequency and loop orientation : for the loop located near the disk center , the optically thin emission ( _ e.g. _ , at @xmath73 ghz ) is sharply peaked near the loop top ; the same effect is seen for the emission near the spectral peak ( at @xmath74 ghz ) , while the optically thick emission ( _ e.g. _ , at @xmath75 ghz ) is weakly peaked at the loop footpoints . for the `` limb a '' orientation ,
the situation is opposite : now the optically thick emission is peaked near the loop top , while the optically thick emission and the emission near the spectral peak are peaked at the loop footpoints . a similar pattern , although with weaker variations , can be seen in the `` limb b '' case . in all cases ,
there are small narrow peaks / dips at the points where the local viewing angle equals @xmath61 ; they are caused by different polarization characteristics of the ordinary and extraordinary electromagnetic modes and are typical of the gyrosynchrotron radiation .
oscillation - induced variations of the radio emission in different parts of the magnetic loop tend to compensate each other , thus reducing the variations of the total ( spatially unresolved ) emission .
due to the variation of the viewing angle ( and hence of the local emission parameters ) along the loop , the mentioned compensation is not complete .
nevertheless , the variations of the total intensity are always much weaker than those of the local ( spatially resolved ) emission ; amplitude of the total intensity variations decreases with a decrease of the relative wavelength of the mhd waves ( _ i.e. _ , with an increase of the number of the waves in the loop ) .
therefore we focus below on the characteristics of the spatially - resolved emission . .
two oscillation phases corresponding to the maximum deviations from the equilibrium are shown .
the asterisks ( @xmath76 ) and diamonds ( @xmath77 ) mark the points with the maximum / minimum magnetic field at the loop axis . ]
figure [ figimages ] demonstrates the simulated radio images at the frequency of 25 ghz ( optically thin emission ) for the standing wave .
two opposite oscillation phases are shown . to make the effect of the mhd waves more pronounced ,
the intensity variations were enhanced as @xmath78 where @xmath79 is the intensity averaged over the wave period @xmath41 : @xmath80 we can see in the figure that for the `` limb b '' orientation ( with an almost constant viewing angle ) the visible emission structure , in general , reflects the structure of the mhd wave : there are three intensity maxima corresponding to the points of maximum magnetic field ( and hence the maximum concentration of energetic electrons ) . for other orientations , due to the projection effects ,
the intensity maxima / minima slightly deviate from the magnetic field maxima / minima ; also , not all mhd waves are visible ( _ i.e. _ , both the intensity and the oscillation - induced variation amplitude depend strongly on the coordinate along the loop ) .
for the propagating mhd waves , the radio images are similar to those shown in figure [ figimages ] , but the brighter / darker spots drift gradually along the loop with time .
we are interested primarily in the qualitative effects that could reveal the basic mhd wave characteristics ( such as the wave mode ) . as mentioned above , in real flaring loops the mhd wave properties ( including the phase speed ) can vary strongly along the loop , which is not considered in our model .
therefore it makes no sense to compare , _
e.g. _ , the simulated time profiles of the emission at the loop top and the footpoints .
instead we consider the local emission parameters , _
i.e. _ , the temporal delays ( phase differences ) between the emissions at different frequencies at a given point . at the loop axis
_ vs. _ the visible coordinate along the loop axis and the oscillation phase . the low - density model ( standing wave )
is used ; the loop contains three sausage waves and the electron spectral index is @xmath72 . for each emission frequency and loop orientation , the maximum modulation depth @xmath81 is given above the corresponding panel . ] , for the propagating wave . ]
figures [ figtds]-[figtdp ] show the time - distance plots of the normalized intensity variation @xmath82 at the axis of the magnetic loop ( _ i.e. _ , along the line formed by the projection of the loop axis on the image plane ) for the low - density model .
four frequencies are shown , representing the optically thick case ( 4 ghz ) , `` moderately '' optically thick case ( @xmath83 ghz ) , spectral peak ( 10 ghz ) , and optically thin case ( 25 ghz ) . above each panel , the respective maximum modulation depth @xmath81 is given ; this value is defined as @xmath84 note that the maximum modulation depth @xmath81 is different from the average modulation depth @xmath85 given by equation ( [ epsilon ] ) ( which will be used in section [ index ] ) . according to figures [ figtds]-[figtdp ] ,
the maximum modulation depth @xmath81 varies from @xmath86 at low frequencies to @xmath87 at high frequencies .
these values are relatively low and , in fact , it would not be easy to detect such oscillations in real observations .
however , this is because we consider the relatively low - amplitude mhd waves : the variations of the emission source parameters @xmath88 and @xmath89 do not exceed @xmath90 and thus the variations of the intensity are comparable to the mhd wave amplitude .
the quasi - periodic pulsations reported by @xcite , @xcite , and @xcite ( with the modulation depths of up to @xmath91 ) require much more powerful mhd waves ; nevertheless , if those pulsations are caused by sausage mhd waves , their characteristics should be qualitatively similar to the case studied in this work .
note also that the intensity variations are more pronounced in the optically thin frequency range , in agreement with the results of @xcite and @xcite .
figure [ figtds ] demonstrates the simulation results for the standing wave .
we can see that the relative modulation depth , in general ( excluding the regions with @xmath71 ) , follows the same trends as the unperturbed intensity ( see figure [ figloop ] ) : for the loop near the disk center , the modulation depth increases towards the loop top for the optically thin emission ( @xmath92 ghz ) and increases towards the loop footpoints for the optically thick emission ( @xmath93 ghz ) ; for the loops near the limb , the trends are opposite .
oscillations at low ( @xmath94 ghz , the first row of the figure ) and high ( @xmath92 ghz , third and fourth rows ) frequencies are synchronous .
the fact that the optically thin and thick emissions oscillate in phase agrees with the results of @xcite , but contradicts the conclusions of @xcite .
the reason is that , in contrast to @xcite , we take into account the modulation of the viewing angle by the sausage wave but neglect the modulation of the energy of the nonthermal electrons . in addition , we use the spatially inhomogeneous model in which the source parameters vary along the line of sight ; this is especially important for the optically thick emission .
since two different approaches ( ours and that of @xcite ) provide opposite results , we conclude that the sausage - mode mhd waves ( at least , in application to the radio intensity modulation ) can not be treated without a spatially resolved 3d model .
note that this behaviour ( in - phase variations of the optically thin and thick emissions ) is observed for all loop orientations and at all locations along the loop ; _ i.e. _ , it is independent of the viewing angle ( except the @xmath95 case which is discussed below ) .
an interesting feature is observed at `` intermediate '' frequencies ( @xmath96 ghz , the second row of figure [ figtds ] ) , _ i.e. _ , in the optically thick range , but not far from the spectral peak ; the optical depth @xmath97 here exceeds unity , but is not very large ( @xmath98 ) .
we can see that the intensity oscillations at these frequencies are shifted by about @xmath16 wavelength with respect to the oscillations at higher and lower frequencies .
this is caused by the wave - induced variations of the viewing angle : at the intermediate frequencies , these variations ( albeit very small ) become the main factor modulating the intensity .
in contrast , at higher or lower frequencies ( with @xmath99 or @xmath100 ) , the effect of changing the magnetic field and electron concentration dominates . at the same time , the modulation depth at the intermediate frequencies reaches its minimum ( in comparison with other frequency ranges ) .
the oscillating magnetic field direction ( and hence the oscillating viewing angle ) also produces some peculiarities in the parts of the loop where the viewing angle is close to @xmath61 . changing the sign of the magnetic field projection on the line of sight results in changing the dominating electromagnetic wave mode which affects the intensity . as a result , in the optically thick frequency range
the mhd wave induces intensity variations of opposite signs at different sides of the @xmath95 stripe ; the modulation depth increases sharply .
in the optically thin frequency range the intensity variations at different sides of the @xmath95 stripe have the same sign and the modulation depth demonstrates only minor variations .
it should be noted , however , that the mentioned peculiarities occur only in very narrow stripes and thus their detection in real observations requires very high spatial resolution .
similar peculiarities ( although without significant changes of the modulation depth ) can be seen for the `` limb a '' orientation , in the optically thick frequency range , at the distance of about @xmath101 from the upper footpoint .
they occur due to the specific loop topology : the optically thick emission is produced near the loop boundary and , for the `` limb a '' orientation ( see figure [ figloop ] ) , the dependence of the viewing angle at the loop boundary surface on the coordinate @xmath58 can become discontinuous ; for other loop orientations , this effect is not observed .
figure [ figtdp ] demonstrates the simulation results for the propagating wave .
since the structure of the standing and propagating sausage waves is the same , all the above conclusions about the radio intensity oscillations remain valid ; the maximum modulation depths shown in figure [ figtdp ] are slightly higher than in figure [ figtds ] . using figure [ figtdp ] , we can track propagation of the mhd waves and estimate their speed . for the low - density model considered here ( see table [ tabmhd ] ) ,
the phase speed of the sausage waves equals 8070 km @xmath102 ; the visible propagation speed of the radio brightenings / darkenings for the `` limb b '' orientation is very close to this value .
however , due to the varying viewing angle , the projected ( visible ) wave speed varies along the loop ; this effect is especially important for the `` disk center '' and `` limb a '' orientations .
the amplitude of the propagating radio intensity variations also varies along the loop with the most notable effects occurring near the regions of transverse ( with respect to the line of sight ) magnetic field : in these regions , the bright / dark stripes in figure [ figtdp ] become discontinuous , although the mhd wave itself has no peculiarities ; the bright / dark stripes in the time - distance plots ( in the optically thick frequency range ) can also become discontinuous in the regions where the viewing angle at the loop boundary sharply depends on the coordinates in the image plane .
thus when interpreting the radio observations with quasi - periodic pulsations , a special attention should be paid to the @xmath95 layers and/or the regions where the visible loop boundary demonstrates topological peculiarities , since they can introduce additional phase shifts which do not actually reflect the properties of the underlying mhd waves .
assuming that the relative intensity variations are approximately proportional to the relative mhd wave amplitude , we can estimate that the mhd waves with the amplitude @xmath88 of about 40 - 50% ( with other source parameters being the same as in figures [ figtds]-[figtdp ] ) should produce the intensity modulation depth of up to 100% .
however , simulating the mhd waves with such amplitudes requires taking into account the nonlinear effects . as will be shown in sections [ index]-[razin ] , the emission modulation depth becomes higher for softer electron beams or for higher plasma densities at low frequencies ( when the razin effect is important ) , although this increase of the modulation depth is always accompanied by a sharp decrease of the intensity .
we now consider how the fluctuations of the radio emission depend on various parameters of the flaring loop .
-[figtdp ] ) at the frequency of 25 ghz for different loop orientations and different numbers of wavelengths ( four and six ) in the loop .
the low - density model ( propagating wave ) is used and the electron spectral index is @xmath72 . ] figure [ figwp ] demonstrates the time - distance plots of the intensity variation for the models with increased numbers of mhd waves in the magnetic loop ( four and six waves , which corresponds to the loop lengths of 11300 and 16950 km , respectively ) ; the simulation results shown correspond to the low - density model , nonthermal electrons spectral index of @xmath72 , and the optically thin emission ( at the frequency of 25 ghz ) . as already shown above ,
the standing and propagating mhd waves affect the radio emission in a qualitatively similar way ; therefore only the results for the propagating wave are shown .
our simulations have revealed no significant effect of the number of the mhd waves in the loop , _ i.e. _ , both the emission modulation depth and the phase relations between the emissions at different frequencies remain the same .
the only difference can be noticed in the regions where the loop axis becomes nearly parallel to the line of sight , _
i.e. _ , near the loop footpoints for the `` disk center '' orientation and near the loop top for the `` limb a '' orientation : since the length of the formation region of the optically thin emission can exceed an mhd wavelength , the effects of variation of the magnetic field and electron concentration in different parts of the emission source ( with opposite phases ) can compensate each other . as a result
, the amplitude of the intensity oscillations decreases with a decrease of the relative ( with respect to the loop length ) mhd wavelength .
this effect is not observed in the optically thick frequency range .
it can be seen in figure [ figwp ] ( as well as in figures [ figimages]-[figtdp ] ) that the mhd observations are best resolved when the magnetic tube is observed in a nearly perpendicular direction . indeed , for the `` limb b '' orientation we can see one - to - one correspondence between the mhd waves and the spatial and temporal structure of the radio emission .
for other orientations , this correspondence is less straightforward and some mhd wave nodes can even be missing in the radio emission ; as expected , the radio emission can not accurately reflect the mhd wave structure in the regions where the viewing angle with respect to the magnetic field @xmath59 is small .
while it is not easy to estimate the number of observed mhd wave nodes in a general case , we can expect ( and this is confirmed by the simulations ) that the mhd wave nodes become missing in the radio emission in the regions with @xmath103 , where @xmath23 is the loop radius and @xmath40 is the mhd wavelength ; this happens either due to the averaging along the line of sight ( in the optically thin frequency range ) or due to the strong reabsorption of the emission during propagation ( in the optically thick frequency range ) .
note also that , as has been discussed above , the @xmath95 layers can introduce the patterns in the radio emission ( at low frequencies ) similar to additional mhd waves .
the energy spectrum of the radio - emitting electrons affects greatly the gyrosynchrotron emission . as can be seen from figure [ figspectrum ] , softer electron beams ( with larger spectral indices @xmath104 ) produce less intense emission with steeper spectrum slope in the optically thin frequency range ; the spectral peak shifts to lower frequencies .
we have found that variation of the spectral index @xmath104 does not change the simulated radio images ( like in figure [ figimages ] ) qualitatively . however , there are quantitative changes in the emission time profiles which may affect detectability of the mhd oscillations .
figure [ figindex ] demonstrates the average modulation depths of the intensity ( for the propagating mhd wave , three sausage waves per loop , low - density model , and `` limb b '' orientation ) ; the emission at the loop axis is considered and the values are plotted as functions of the coordinate along the loop .
the average modulation depth @xmath85 is calculated as @xcite @xmath105 ^ 2\mathrm{d}t;\ ] ] it is related to the maximum modulation depth @xmath81 ( shown in figures [ figtds]-[figwp ] ) as @xmath106 .
we can see that in the optically thin frequency range ( figure [ figindex]b ) an increase in the spectral index results firstly in an increase in the modulation depth . as a result ,
the oscillation - modulated radio images become more contrasted and the periodic variations of the emission become easier to detect . for even higher electron spectral indices ( @xmath107 for the considered frequency of 25 ghz ) , the modulation depth decreases sharply , because the gyrosynchrotron emission becomes too weak and the emission is produced mainly by the free - free mechanism that is weakly affected by the mhd oscillations . in the optically thick frequency range
( figure [ figindex]a ) the modulation depths are low ( @xmath108 ) and decrease with increasing @xmath104 .
as mentioned above , figure [ figindex ] corresponds to the propagating wave and `` limb b '' orientation ; for other orientations and/or for the standing wave , the above conclusions remain the same . assuming that the relative intensity variations are approximately proportional to the relative mhd wave amplitude
, we can estimate that for the electron spectral index of @xmath109 and the mhd wave amplitude @xmath88 of about 25 - 30% , the average ( @xmath85 ) and maximum ( @xmath81 ) modulation depths in the optically thin frequency range should reach @xmath110 and @xmath111 , respectively ; the mhd waves with such amplitudes seem to be feasible .
finally , we consider the case when the spectral peak of the gyrosynchrotron radio emission is formed by the razin effect @xcite .
this effect takes place in a relatively high - density plasma and results in a considerable suppression of the gyrosynchrotron emission at the frequencies below @xmath112 where @xmath113 and @xmath114 are the plasma and cyclotron frequencies in the emission source , respectively . at the frequencies below @xmath115 , the variations of the thermal plasma density
should become the main factor affecting the radio intensity . as an illustrative example
, we use the high - density model ( see table [ tabmhd ] ) . as can be seen from figure [ figspectrum ] , in this case
the razin frequency @xmath115 is about 4 ghz .
thus at the frequencies well above 4 ghz the intensity variations should be caused mainly by the oscillations of the magnetic field and energetic electron concentration ( like in the low - density model ) , at the frequencies below 4 ghz the intensity variations should be caused mainly by the razin effect ( _ i.e. _ , by the oscillating plasma density ) , while at the frequencies around the spectral peak ( @xmath75 ghz ) both factors should be important .
[figwp ] ) for the high - density model .
standing wave is considered and the electron spectral index is @xmath72 . ]
figure [ fighd ] demonstrates the simulation results ( time - distance plots of the intensity variation ) at three different frequencies for the high - density model , standing sausage wave , and the nonthermal electrons spectral index of @xmath72 . in the optically thin frequency range , the modulation depth of the intensity is larger than that for the low - density model due to the model parameters used ( the relative mhd wave amplitude for the high - density model is larger , see table [ tabmhd ] ) .
the modulation depth reaches its minimum at a frequency below the spectral peak ( @xmath116 ghz , depending on the viewing angle ) and then rapidly increases with decreasing frequency ; however , as can be seen in figure [ figspectrum ] , the intensity at those frequencies also becomes very low , so that detection of the intensity oscillations in observations would be difficult .
it can be seen in figure [ fighd ] that for the `` limb b '' orientation ( when the viewing angle is close to @xmath61 ) the radio oscillations in the optically thin frequency range ( @xmath92 ghz ) and near the spectral peak ( @xmath75 ghz ) are synchronous ( in phase ) , while the low - frequency emission ( @xmath117 ghz ) oscillates in anti - phase with the emissions at higher frequencies ; the phase reversal occurs ( for @xmath118 ) at @xmath116 ghz .
clearly , this result indicates that at high frequencies ( above and even slightly below the spectral peak ) the intensity variations are caused by the variations of the magnetic field and nonthermal electrons concentration , while at lower frequencies ( @xmath119 ghz ) the razin effect dominates ; these factors affect the emission in opposite ways .
a similar picture can be seen for two other loop orientations in the regions where the viewing angle is close to @xmath61 ( _ i.e. _ , near the loop top for the `` disk center '' orientation and near the footpoints for the `` limb a '' orientation ) . for smaller viewing angles
the picture is more complicated , since both the modulation depth and the above - mentioned frequency of the oscillation phase reversal are angle - dependent : the frequency of the phase reversal increases with a decrease of the viewing angle . as a result ,
if the viewing angle varies gradually along the loop , we can see either correlation or anti - correlation between the oscillations at two given frequencies , depending on the coordinate @xmath58 .
the above results contradict the conclusions of @xcite who have found that the oscillations of the high- and low - frequency emissions ( in the case when the razin effect is important ) should occur strictly in phase .
as mentioned before ( see section [ phase ] ) , this is because we consider a slightly different model for the variation of the parameters in the sausage wave , as well as the spatially inhomogeneous emission source .
our results also disagree with the conclusions of @xcite who have found that the oscillations of the high- and low - frequency emissions occur mostly in phase ( although this conclusion was dependent on the viewing angle ) .
most likely , this is because @xcite considered the emission with the frequencies below the spectral peak but above the `` phase reversal frequency '' so that the effect of the razin suppression was too weak .
we have simulated the modulation of the gyrosynchrotron radio emission by the sausage - mode mhd oscillations in a semi - circular magnetic loop , in application to the solar flares .
although the model used does not include the variation of the magnetic field and plasma parameters with height , it has allowed us to study the effect of varying the viewing angle along the loop .
we have considered three loop orientations and different parameters of the magnetic field , plasma , and energetic particles . in particular , we have studied in detail the phase relations between the spatially resolved intensity oscillations at different frequencies .
the results can be summarized as : * in the model with a relatively low plasma density ( when the razin effect is negligible ) , the high - frequency ( optically thin ) and low - frequency ( optically thick ) emissions oscillate , as a rule , in phase .
an exception is the optically thick emission at the frequencies just below the spectral peak , whose oscillations are shifted by @xmath120 mhd wavelength with respect to the oscillations at other frequencies . * in the model with a relatively high plasma density ( when the razin effect is important ) , the emissions at high and low frequencies oscillate in anti - phase .
the reversal of the oscillation phase occurs , as a rule , at a frequency well below the spectral peak ; this frequency depends on the viewing angle . * in the optically thin frequency range , both the unperturbed intensity and the absolute and relative intensity variations are the highest in the regions where the local magnetic field is nearly perpendicular to the line of sight .
in contrast , in the optically thick frequency range all the mentioned values are the highest in the regions with small viewing angles .
* in the optically thin frequency range , the intensity oscillations are more pronounced for softer electron beams ( _ i.e. _ , with higher spectral indices ) . in the optically thick frequency range
the trend is opposite .
* in the optically thick frequency range , the intensity oscillations can have sharp phase shifts between the adjacent regions due to the presence of layers of transverse ( with respect to the line of sight ) magnetic field and other topological peculiarities . *
averaging over the visible source area and/or along the line of sight reduces the amplitude of the intensity oscillations ; this effect becomes more important with an increase in the number of the mhd harmonics in the loop .
the latter result means that , due to a limited spatial resolution , the real observations can differ from the ideal simulations presented here .
nevertheless , we expect that if an instrument spatial resolution is better than a mhd wavelength , this will allow us to detect ( or rule out ) the above - mentioned phase relations between the oscillations at different frequencies ; another requirement is the ability to perform imaging observations at many frequencies .
we anticipate that the forthcoming multiwavelength radioheliographs ( such as the expanded owens valley solar array , upgraded siberian solar radio telescope , and chinese spectral radioheliograph ; see the review of @xcite and references therein ) will be able to provide the necessary data .
this work was supported in part by the russian foundation of basic research ( grants 12 - 02 - 00173 , 13 - 02 - 90472 and 14 - 02 - 91157 ) and by a marie curie international research staff exchange scheme `` radiosun '' ( people-2011-irses-295272 ) .
the research leading to the presented results has been sponsored by an odysseus grant of the fwo vlaanderen .
it has been performed in the context of the iap p7/08 charm ( belspo ) and the goa-2015 - 014 ( ku leuven ) .
as was mentioned above , we consider the mhd waves in an overdense cylinder ; the system can be described by the cylindrical coordinates : the distance along the cylinder axis @xmath1 , the distance from this axis @xmath2 , and the azimuthal angle @xmath121 ( see figure [ figcoordinates]a ) .
we assume that the cylinder has the length @xmath122 ( _ i.e. _ , @xmath123 ) and the radius @xmath23 ( the radio emission is produced only inside the cylinder , _ i.e. _ , at @xmath124 ) .
the cylinder is assumed to be bent into a semi - circular ( semi - toroidal ) loop ( see figure [ figcoordinates]b ) ; the loop curvature radius is @xmath125 ( evidently , we need @xmath126 ) .
we assume that initially the loop is oriented as shown in figure [ figcoordinates]b , _
i.e. _ , it is vertical , symmetric with respect to the @xmath127 and @xmath128 planes , and its axial line lies in the @xmath127 plane .
then the cylindrical coordinates @xmath129 are related to the cartesian coordinates @xmath130 as @xmath131 and @xmath132 where @xmath133 .
assume that a vector @xmath134 at a given point is defined by its three components @xmath135 in the cylindrical coordinate system .
after transformation to the semi - circular loop , the vector can be expressed as @xmath136 , where the unit vectors @xmath137 in the cartesian coordinates are given by @xmath138 the above formulae are valid in a general case . in this work
we consider the azimuthally symmetric case when all parameters are independent of the azimuthal angle @xmath121 and the azimuthal components @xmath139 of all vectors are equal to zero .
the orientation shown in figure [ figcoordinates]b corresponds to the vertical loop located at the center of the solar disk , with the axis @xmath140 being the line of sight and the plane @xmath127 being the equatorial plane .
all other orientations ( see , _
e.g. _ , figure [ figloop ] ) are obtained by rotating the loop firstly by the angle @xmath57 around the local vertical ( @xmath140 axis ) and then by the angle @xmath40 around the solar rotation axis .
59 # 1isbn # 1#1#1#1#1#1#1#1_#1_#1*#1*#1#1#1#1#1#1#1#1#1#1#2#1#1#1#1#1_#1_#1#1#1*#1*#1#1#1_#1_#1#1#1#1#1#1#1#1#1#1#1 # 1http://dx.doi.org/#1 [ ] # 1http://arxiv.org / abs/#1 [ ] # 1http://adsabs.harvard.edu / abs/#1[]#1#1#1#1#1#1#1#1#1#1 # 1#1 | radio observations of solar flares often reveal various periodic or quasi - periodic oscillations .
most likely , these oscillations are caused by magnetohydrodynamic ( mhd ) oscillations of flaring loops which modulate the emission .
interpretation of the observations requires comparing them with simulations .
we simulate the gyrosynchrotron radio emission from a semi - circular ( toroidal - shaped ) magnetic loop containing sausage - mode mhd oscillations .
the aim is to detect the observable signatures specific to the considered mhd mode and to study their dependence on the various source parameters .
the mhd waves are simulated using a linear three - dimensional model of a magnetized plasma cylinder ; both standing and propagating waves are considered .
the curved loop is formed by replicating the mhd solutions along the plasma cylinder and bending the cylinder ; this model allows us to study the effect of varying the viewing angle along the loop .
the radio emission is simulated using a three - dimensional model and its spatial and temporal variations are analyzed .
we consider several loop orientations and different parameters of the magnetic field , plasma , and energetic electrons in the loop . in the model with low plasma density ,
the intensity oscillations at all frequencies are synchronous ( with the exception of a narrow spectral region below the spectral peak ) . in the model with high plasma density , the emission at low frequencies ( where the razin effect is important ) oscillates in anti -
phase with the emissions at higher frequencies .
the oscillations at high and low frequencies are more pronounced in different parts of the loop ( depending on the loop orientation ) .
the layers where the line - of - sight component of the magnetic field changes sign can produce additional peculiarities in the oscillation patterns . |
black hole x - ray transients ( bhxrts ) , also known as x - ray novae and soft x - ray transients , are low - mass x - ray binaries ( lmxbs ) which undergo dramatic x - ray , optical , and radio outbursts , separated by years , decades or even longer periods of quiescence ( @xcite;@xcite ) .
the class is diverse , with orbital periods of 4hrs to a few days to a month .
outburst behavior is similarly varied . among these , however , are a sub - group which show relatively consistent behavior and these are thought of as the ` typical ' members of the class . they are often compared to the prototypical bhxrt , a062000 ( see @xcite for a review ) . in such a typical outburst , the x - ray emission is usually ( but not always , see @xcite ) dominated by thermal emission from the hot inner accretion disk , and uv and optical emission is thought to be produced by reprocessing of x - rays by the outer disk .
a number of attempts have been made to characterize the spectral energy distributions ( seds ) of these bhxrts in outburst . @xcite and @xcite attempted to use simple disk spectral models , and other authors
have used a variety of black bodies and power - laws .
these mostly only consider a single source at a time , use a variety of extinction curves and reddening determination methods , and do not use comparable spectral models .
consequently it is difficult to compare the data on different sources and attempt to draw broad conclusions about the population as a whole .
the goal of this paper is to bring together all of the useful data on a suitable homogeneous sample of sources , with recalibrations where appropriate for the satellite data .
it is then possible to deredden the spectra consistently and draw direct empirical conclusions about the observed seds .
we can also go on to compare the complete sample with simple models in a consistent way .
finally we can hope that this compilation will provide a reference set as more sophisticated disk spectral models are developed .
llrccc object & outburst peak & @xmath4 & adopted & dust map & distance + & ( approx . ) & ( hrs ) & @xmath5 & @xmath5 & ( kpc ) + a 062000 = v616 mon & 1975 aug 13 ( mjd 42637 ) & 7.8 & @xmath6 & 0.49 & @xmath7 + gs 1124684 = gu mus & 1991 jan 15 ( mjd 48271 ) & 10.4 & @xmath8 & 0.37 & @xmath9 + gro j0422 + 32 = v518 per & 1992 aug 15 ( mjd 48848 ) & 5.1 & @xmath10 & 0.36 & @xmath11 + grs 100945 = mm vel & 1993 sep 15 ( mjd 49245 ) & 6.9 & @xmath12 & 0.19 & @xmath13 + grs 1716249 = v2293 oph & 1993 sep 30 ( mjd 49260)&@xmath1414.7 & @xmath15 & 1.08 & ( @xmath16 ) + xte j1859 + 226 = v406 vul & 1999 oct 16
( mjd 51467 ) & 9.1 & @xmath17 & 0.49 & ( @xmath18 ) + studies of these broad - band seds are important as they can inform us about the temperature structure within the binary .
they are especially useful for the typical bhxrt systems considered , as these mostly have very small mass ratios , hence the emission from reprocessing on the companion star is weak , and the systems so far discovered have intermediate or low inclinations , so disk self obscuration and limb darkening should not be large effects . in these cases
the comparison between observed sed and disk temperature structure should be relatively direct .
one must of course assume an emergent spectrum for the disk . at present ,
black bodies are probably as good an approximation as any for optically thick disks in x - ray binaries , but we can anticipate models including a more rigorous treatment of the disk atmosphere in the future .
the observed sed essentially can tell us the distribution of emitting area as a function of temperature . given assumptions such as axisymmetry and a monotonic dependence of temperature on radius
then one can map this area temperature relation into the radial temperature distribution , but the more generalized statement is useful in considering more complex geometries which could be imagined , such as disks which are warped or exhibit spiral waves . in this more general case ,
given a three - dimensional disk geometry , a model for the irradiation geometry ( e.g. an isotropically emitting point source , or perhaps a flat central disk ) , and some method to calculate emergent spectra ( at its simplest , a black body ) , one can test specific models against the observed spectra , and falsify some combinations .
there is certainly more than enough uncertainty about the temperature structure of irradiated disks to warrant seeking further observational constraints .
for example , theoretical work suggests that disks should be self - shielding ( @xcite ; @xcite ; @xcite ) , and hence that irradiation requires scattering of x - rays or warping of the disk @xcite .
these two cases may produce very different temperature structures .
an understanding of irradiation geometry has immediate application in providing input to numerical models of bhxrt outbursts , as these models have to make some assumptions about the strength , and radial dependence , of x - ray heating ( e.g. @xcite ; @xcite ; @xcite ; @xcite ) .
we begin in section [ selectionsection ] by defining our sample of bhxrts . in section [ reddeningsection ]
we review and evaluate reddening determinations for these sources , in particular incorporating a reanalysis of 2175 bump data for some sources .
this section also discusses the importance of the extinction curve adopted .
section [ distancesection ] discusses corresponding distance estimates .
we then proceed to review the database of spectra available based on satellite observations ( mainly uv ) in section [ satellitesection ] , and ground - based observations ( optical ) in section [ opticalsection ] .
having collated our sample spectra , we then compare the optical spectra in section [ opticalsedsection ] and the broader - band optical - uv spectra in section [ uvsedsection ] .
we next take an interlude to evaluate the effect of reddening uncertainties upon the spectra in section [ dereddenvarsection ] before comparing the spectra with disk models in section [ modelsection ] .
in section [ discussionsection ] we compare our results with other studies and with theoretical work . finally in section [ conclusionsection ]
we summarize our conclusions and consider where progress can most effectively be made in the future .
we restrict our sample of sources in order to obtain a relatively homogeneous set of spectra .
our selection criteria are as follows .
firstly we consider only sources which are believed to harbor a black hole . in most cases
this selection is based on a dynamical mass estimate , but where this is not available , more subjective criteria are used . any source exhibiting type i x - ray bursts
is obviously excluded .
secondly we exclude those sources with an orbital period greater than one day .
most long period objects exhibit unusual outbursts , several have companions luminous enough to contribute even during outburst and the large disks expected in these systems may well behave rather differently to those in the short period systems .
finally we only include high luminosity systems showing a ` canonical ' lightcurve , i.e. one with a fast rise and quasi - exponential decay .
specifically we exclude xte j1118 + 480 ; this seems appropriate as this source exhibited an unusually faint outburst and the optical light probably includes significant non - disk emission , likely of a synchrotron origin ( e.g. @xcite )
. for the sources included , only spectra from the exponential decay phase of the lightcurve are used , i.e. we exclude mini - outburst spectra .
an additional practical criterion is of course that there must exist flux - calibrated spectra .
based on these criteria we obtain a primary sample of sources with relatively complete optical and uv coverage and a secondary sample with only optical spectra .
our primary sample includes a 062000 , gs 1124684 , gro j0422 + 32 , and xtej1859 + 226 .
our secondary sample adds grs 100945 , and grs 1716249 .
we initially included gs 2000 + 25 in the secondary sample , as a fluxed spectrum was presented by @xcite and it otherwise meets our selection criteria .
when we came to estimate the reddening , however , it was clear that it was too large , and too uncertain , for a meaningful derivation of the intrinsic spectral energy distribution , so we have not attempted an analysis of this spectrum . for all of the other targets selected @xmath19 and is relatively well determined .
all of these objects also lie at least @xmath20 from the galactic plane . for the sources considered ,
the optical and uv emission should be dominated by the disk , although the ir may in some cases include a synchrotron component ( e.g. @xcite ; @xcite ) , so we do not include ir coverage .
since most bhxrts lie at significant distances , they are typically rather reddened objects , hence a reliable correction for interstellar reddening is crucial if we are to accurately reconstruct the spectral energy distribution .
this correction depends upon two things : the determination of the reddening value , typically parameterized in @xmath5 , and an extinction curve specifying the amount of extinction as a function of wavelength . for a review of the problem see @xcite .
a number of methods are available to estimate the reddening , but they vary considerably in their reliability .
the most direct measures are not necessarily the best ; these include the optical color ( e.g.@xmath21 ) or emission line ratios .
these do directly measure the differential extinction as a function of wavelength , but their interpretation is model dependent .
less direct methods can be more independent of assumptions about the target , but instead involve assumed properties for the line - of - sight material , for example the gas to dust ratio , the abundance of nai , or the strength of the 2175 absorption feature relative to the broader structure in the extinction curve .
several approaches also exist for choosing an extinction curve , as discussed by @xcite . as we typically have no independent information on the properties of the interstellar medium toward a bhxrt
, we will adopt the @xcite @xmath22 extinction curve as this should be the ` best ' current galactic average curve . in analyzing our results
we must then remember that our uncertainties depend not only on the uncertain amount of extinction but the uncertain shape of the extinction curve .
our preferred method to estimate the amount of extinction is to measure the strength of the 2175 interstellar absorption feature . in fitting the feature
we must assume some underlying spectral shape , either fixed or with some free parameters , and then adjust @xmath23 to fit the data .
it should be borne in mind that the strength of the feature is one of the variable characteristics of the uv extinction curve ; hence the correlation between the 2175feature and @xmath5 is not perfect , and the intrinsic variance in their relative values introduces an unavoidable 20percent error in reddenings determined in this way @xcite .
this uncertainty can only be avoided if the relative strength of the feature for the line of sight is known independently from nearby stars , which is only the case for one of our sources , a 062000 @xcite . in spite of these limitations ,
the method is generally more precise than the alternatives .
the equivalent width of the nad interstellar doublet can in principle be used to estimate the reddening , and calibrations exist for this purpose .
the problem is that _ several _ calibrations exist and these do not agree . @xcite
derive @xmath24 from observations of supernovae in external galaxies .
@xcite on the other hand obtained @xmath25 by analyzing the data of @xcite on southern supergiants .
furthermore , as noted by @xcite , the nad lines are not in general sensitive to @xmath26 due to saturation , so this method can underestimate the reddening ; whether it does or not depends on the detailed substructure of the lines which is not usually resolved .
some caution should therefore be exercised in applying this method .
in addition to the nad lines , relations between the equivalent widths of diffuse interstellar bands and @xmath5 are commonly used .
@xcite performed an extensive study and tabulated relations for 17 interstellar bands in the 44006850 range .
the correlation between different lines was found to be very good but there were significant regional effects .
another approach taken is to adopt scaling ratios between interstellar gas and dust densities , usually expressed in a regression relation between @xmath27 and @xmath28 or @xmath5 .
this method is plagued with uncertainty , both due to intrinsic variance in this ratio , and due to different measures of @xmath27 ; it may be measured directly though uv ly@xmath29 absorption ( e.g. @xcite ; @xcite ) or indirectly through soft x - ray absorption which also includes metallic absorption ( e.g.@xcite ; @xcite ) ; for the latter it is necessary to assume abundances .
we will adopt the relation of @xcite : @xmath30@xmath31 , as this relation is relatively direct and conveniently falls intermediate between the other estimates referred to .
these authors note that individual sources typically scatter by 30percent around this line , and adopting this uncertainty encompasses the other average relationships as well .
an additional concern for measurements of n@xmath32 is the possible presence of local absorption intrinsic to the source .
the local absorbing material will probably be too hot to include a dusty component , so will not contribute to the optical / uv absorption .
some authors assume a typical intrinsic @xmath33 color in outburst , and hence derive a direct estimate of @xmath5 . for our purposes , however , this assumption is extremely dangerous ; to use such estimates without independent corroboration would lead to a circular logic .
we would be assuming that the intrinsic spectra of all bhxrts are the same to determine their reddening and hence derive their intrinsic spectrum . a direct measure of the reddening of the spectrum
can alternatively be obtained from the balmer decrement , by assuming line ratios appropriate for case b recombination ( e.g.@xcite ) .
this approach is obviously dependent upon the assumption that case b line ratios are appropriate , and so is also not a very secure method .
a final technique , of more use for extragalactic work , is to use maps of galactic dust , for example @xcite .
the problem with applying this technique to galactic objects is that it only yields the total amount of reddening along that line of sight , some of which may be behind the target .
this technique is thus mainly useful as an upper limit , although some sources such as gro j0422 + 32 are believed to lie above the dusty component of the galactic disk .
an additional concern is that these maps do not resolve fine extinction structure and so are only reliable for sources above or below the galactic plane ; @xcite recommend only trusting @xmath34 .
fortunately , the nature of our sample is such that highly reddened objects in the plane have been excluded , and all of our targets happen to lie outside of this band .
we therefore apply this technique , remembering that it may include background extinction . in most cases
the reddening estimated in this way is comparable to , or somewhat larger than , other methods so it does seem to be of use , at least for high latitude x - ray binaries .
@xcite derived @xmath35 for a 062000 from fitting _
ans _ uv data through 2175 feature .
subsequently , @xcite used spectra from 25 nearby early type stars to derive a line - of - sight extinction curve for a 062000 .
it was found that the 2175 feature was rather stronger along this line - of - sight than average , and hence they derived a revised lower estimate of @xmath36 .
the uncertainty in this estimate will be similar to the earlier measurement ; since they used a line - of - sight extinction curve there is no need to account for a 20percent error due to uncertainty in the intrinsic 2175 strength . @xcite
estimated @xmath37 , with a large uncertainty , from the nad lines .
they note , however , that the dibs seem unusually strong for the reddening derived .
@xcite collated the dib measurements and confirmed this conclusion .
they emphasize , however , that the strength of dibs as a function of @xmath5 does vary with galactic longitude , and that a 062000 lies within @xmath20 of the region where the dibs are relatively strongest .
using appropriate relations for this longitude they derive @xmath38 or @xmath39 , depending on the relation used .
the reddening derived from optical interstellar features is therefore broadly consistent with that obtained from the uv . both are lower than the value of 0.49 derived from dust maps @xcite but a 062000 is the closest of our objects and background extinction is expected .
we adopt the _ ans _
value of @xmath6 for the remainder of this work .
gs 1124684 was among the best studied in the uv with a series of _
iue_observations @xcite early in the outburst and a later , higher signal - to - noise _ hst _ one @xcite .
these permit fitting of the 2175 feature to derive @xmath5 . @xcite
derive @xmath40 using this method and @xcite obtain @xmath41 from considerably higher signal - to - noise data .
both use the @xcite extinction law . repeating the analysis of the _ hst _ data using the @xcite extinction curve , and a more realistic estimate of the accuracy of the method , we derive @xmath42 from these data in section [ hstsection ] , consistent with both of the earlier estimates .
from interstellar absorption lines , @xcite obtain two values : 0.3 from applying the calibration of @xcite to the nad lines ( with a measured equivalent width of 1.4 ) and 0.35 using the calibration of @xcite on the 5778 , 5780 , 5797 absorption bands .
these are consistent with the uv determination .
the hydrogen column density has been estimated from _ ginga _ data alone as about @xmath43@xmath44 @xcite , and from _
ginga_and _ rosat _
spectra at @xmath45@xmath44 @xcite .
these imply reddenings of @xmath46 and 0.4 respectively , which are just about consistent with optical estimates given the uncertainty in the relation between the two .
for the remainder of this work we will adopt our revised estimate based on the 2175 feature : @xmath42 .
as for a 062000 , this is slightly lower than was derived from dust maps , 0.37 .
the 2175 feature in gro j0422 + 32 has been measured by @xcite using _ iue _ data .
they obtain @xmath47 .
we have repeated the analysis using a more recent extinction curve and obtain a range of values of 0.290.48 , depending on the spectral range modeled and the underlying spectrum which is assumed ( see section [ iuesection ] ) .
many authors have also attempted to derive a reddening from interstellar lines . @xcite
suggest @xmath48 based on 4428 and 6613 dibs .
@xcite also obtain @xmath49 from the 5780 dib .
finally @xcite measure equivalent widths of @xmath50 for the nad lines and @xmath51 for the 5780/5788/5797 group . using the relations of @xcite and
@xcite they deduce @xmath52 .
@xcite examine 12 stars in the field surrounding gro j0422 + 32 and find that their color - color diagram needs to be dereddened by @xmath53 in order to match the unreddened main sequence .
this method assumes that all these stars , and gro j0422 + 32 , are subject to the same reddening ; it would be appropriate if all are above the main plane of extinction .
even if some are not , but gro j0422 + 32 is ( as seems likely given the estimated distance for gro j0422 + 32 ) , then this value becomes a lower limit .
@xcite fit absorbed power - laws to _ asca _ spectra from the 1993 august mini - outburst to obtain @xmath54@xmath44 , corresponding to @xmath55 .
finally , and most recently , @xcite derive @xmath56 from fitting the _ quiescent _ spectral energy distribution , but this value is somewhat sensitive to assumptions about the amount and color of disk light .
there is clearly little close consensus on the reddening value , and even the method which is usually most precise , fitting the 2175 feature , yields equivocal results .
consequently , based on the spread in measured values we adopt @xmath57 .
this choice is rather arbitrary , but is representative of the scatter between the derived estimates .
it is completely consistent with the estimate of 0.36 from dust maps .
@xcite present several arguments for a low reddening in grs 100945 , @xmath58 . from the strength of nad absorption they derive an estimate of @xmath59 , with an upper limit of @xmath60 .
assuming that the intrinsic @xmath33 color is similar to other lmxbs implies a consistent value of @xmath61 ( although see caution above ) .
we follow these authors in adopting @xmath62 .
the column density is not well constrained , but probably @xmath63@xmath44 @xcite , which implies @xmath64 , consistent with the optical estimates .
finally dust maps predict @xmath65 , consistent with other estimates .
the reddening of grs 1716249 has been estimated by @xcite as @xmath66 based on a number of arguments giving roughly consistent values .
the strength of the nad lines implies @xmath67 ( see the note on saturation of these lines above ) .
the strength of dibs suggests @xmath680.9 .
arguments based on the outburst @xmath33 color , and the balmer decrement give somewhat higher values , but are model dependent as discussed above .
the interstellar hydrogen column density has been estimated at @xmath69@xmath44 ( tanaka 1993 ) , implying @xmath70@xmath71 , which is consistent with the optical estimates .
finally , dust maps imply @xmath72 . in summary , we see no compelling reason to amend the estimate of @xcite , leaving this as the largest and most uncertain reddening in our sample .
@xcite estimated @xmath73 using the well defined 2175 interstellar feature .
@xcite obtained @xmath74 from the 5780 dib .
estimates of the hydrogen column density included @xmath75@xmath76 ( @xcite from _ sax_data ) and @xmath77@xmath44 ( @xcite ; _ rxte _ ) .
this column density corresponds to @xmath78@xmath79 , allowing for the uncertainty and variance in the gas - dust ratio . even with this uncertainty , and that in the 2175determination , there is then a discrepancy .
in contrast , dust maps imply an even lower value , @xmath80 .
the 2175 value thus represents a reasonable middle - ground , so we will use this estimate , and the derived dereddened spectra from @xcite .
the determination of distances to bhxrts is even more problematic than their reddenings , since there are no direct measures of distance .
ideally , we would seek some kind of standard candle to determine the distance . for neutron star lmxbs
, a standard candle does exist in the form of radius expansion type - i x - ray bursts .
no such analog exists for black holes , and any attempt to use the outburst luminosity , as done by some authors , runs into a circular logic problem again , that we would be artificially forcing uniformity on our sample before comparing them .
the only reasonably robust method that has been applied to our sample is to use the quiescent companion star .
system parameters provide estimates of the companion s temperature and radius , and hence absolute magnitude , albeit with a large uncertainty in many cases .
given the observed magnitude , estimated extinction and the fraction of quiescent light due to the disk ( the veiling ) we can then derive a distance .
the errors are likely to be large , but are representative of the real uncertainty in knowing the distance , and other methods are likely to be even worse .
the best method to assess the errors is to perform a simple simulation , constructing a population of model binaries consistent to within the observational uncertainties and derive a distance for each one , and hence a mean distance and standard deviation .
this method has recently been applied to a number of our sources using ir data .
this is preferable because the ir derived distance is less sensitive to uncertainty in extinction or companion temperature than the optical is .
the best estimate is for a 062000 for which @xcite derive @xmath7kpc . for gs 1124684 @xcite quote @xmath81kpc with no uncertainty , but this estimate was subsequently refined with simulations as above to @xmath9kpc @xcite . for gro
j0422 + 32 @xcite estimate @xmath11kpc and finally for grs 100945 @xcite derive @xmath13kpc .
of course , these are not the only estimates for these sources .
however we believe they are the most thorough and reflect the most up to date parameter estimates .
these estimates are broadly consistent with previous values .
some concerns do still exist for these objects in that these estimates , and their quoted uncertainties , depend on the limits placed upon the veiling .
the works cited generally assume rather small veilings in the ir , and while this assumption is widely made , its validity is questionable @xcite . the most immediate effect of increasing the veiling is to reduce the apparent brightness of the companion star , implying a larger distance .
it is therefore quite possible that the distances are underestimated somewhat . for more detailed discussion of these issues
see @xcite . for the other two of our selected sources ,
the information necessary to use the quiescent companion as a standard candle is simply not all available yet , so in these cases we make use of other indicators from the literature .
the distance to grs 1716249 is the poorest determined in our sample .
@xcite estimate 2kpc from relations between nad equivalent width and distance .
since the nad lines could well be saturated for the relatively high extinction of this object , this distance is really only a lower limit , even assuming that the mean relation is valid along this line of sight .
based on the peak optical brightness , these authors estimate an upper limit of 2.8kpc , but as noted above we should be cautious about using such measurements for our purposes . in this case , however , we have little other information to go on and follow @xcite in adopting @xmath82kpc for grs 1716249 .
the true uncertainty may well be larger , however , and the distance could be significantly greater .
the distance to xte j1859 + 226 is also problematic .
@xcite estimated @xmath83kpc from both the brightness of the outburst and that of the quiescent companion .
the latter estimate was , however , based largely upon guesses for a number of binary parameters including the orbital period and companion spectral type .
@xcite report spectroscopy indicating an orbital period longer than assumed by @xcite and an earlier spectral type .
both effects will tend to make the quiescent companion more luminous and hence push the source to a larger distance .
we have done a simulation similar to those discussed above to test these effects , and estimate that the quiescent brightness implies a distance of @xmath84kpc assuming the constraints of @xcite .
this value depends on some uncertain or totally unknown parameters , however , such as the inclination and mass ratio .
the disk veiling is also ignored and a significant disk contribution would push the distance even higher .
the faintness in quiescence therefore suggests a very large distance , large enough to conflict with other estimates .
outburst x - ray observations place a likely constraint of 513kpc , based on a combination of spectral and timing information @xcite .
fits to the optical uv spectral energy distribution suggested 4.68.0kpc @xcite , assuming a plausible range of possible disk radii and inclinations . in view of the extreme uncertainty about the system parameters , in particular the companion star s effective temperature , we believe these outburst estimates are currently more reliable , but that we have no really trustworthy distance for this object . we adopt a working estimate of @xmath18kpc , which encompasses most of the parameter space proposed thus far , but
note that a larger distance remains a possibility .
llrllcl object & date & outburst & facility & wavelength & plotted & reference + & & day & & range ( ) & & + a 062000 & 1975 sep 2830 & 4648 & ans & 15503300 & y & 1 + gs 1124684 & 1991 jan 1721 & 26 & iue & 11503300 & a & 2 + & 1991 jan 24 & 9 & iue & 11503300 & a & 2 + & 1991 jan 29 & 14 & iue & 11503300 & n & 2 + & 1991 feb 6 & 22 & iue & 11503300 & a & 2 + & 1991 feb 12 & 28 & iue & 11503300 & a & 2 + & 1991 feb 18 & 34 & iue & 11503300 & a & 2 + & 1991 feb 25 & 41 & iue & 11503300 & a & 2 + & 1991 feb 28 & 44 & iue & 11503300 & a & 2 + & 1991 mar 15 & 59 & iue & 11503300 & n & 2 + & 1991 mar 2728 & 7172 & iue & 11503300 & n & 2 + & 1991 apr 22 & 97 & iue & 11503300 & n & 2 + & 1991 may 15 & 120 & hst / fos & 15804810 & y & 3 + gro j0422 + 32 & 1992 aug 16 & 2 & iue & 11503300 & a & 4 + & 1992 aug 20 & 6 & iue & 11503300 & a & 4 + & 1992 aug 26 & 10 & iue & 11503300 & a & 4 + & 1992 sep 2 & 19 & iue & 11503300 & a & 4 + & 1992 sep 10 & 27 & iue & 11503300 & a & 4 + & 1992 sep 20 & 37 & iue & 11503300 & a & 4 + & 1992 sep 27 & 44 & iue & 11503300 & a & 4 + xte j1859 + 226 & 1999 oct 18 & 2 & hst / stis & 1150 - 10200 & y & 5 + & 1999 oct 27 & 11 & hst / stis & 1150 - 10200 & y & 5 + & 1999 nov 6 & 21 & hst / stis & 1150 - 10200 & y & 5 + & 2000 feb 8 & 115 & hst / stis & 1150 - 10200 & y & 5 + & 2000 mar 5 & 141 & hst / stis & 1150 - 10200 & y & 5 + a 062000 was observed by the ultraviolet experiment onboard the astronomical netherlands satellite ( ans ) around day 47 after x - ray peak @xcite ; a second observation was made around day 225 @xcite which we do not use as it was obtained beyond the main exponential decay phase of the outburst .
@xcite tabulate the data from the earlier observation . from these we constructed an average using only the six observations when the broad 1550 bandpass was used to minimize the impact of the possible civ emission line . _
iue _ spectra were obtained for two of our sources : gs 1124684@xcite and gro j0422 + 32 @xcite . both sources
were observed with the low - resolution short- and long - wavelength prime cameras ( swp and lwp ) .
we obtained all of these data from the archive , using the newsips reprocessed results @xcite . for gs 1124684
, good coverage was obtained throughout the outburst , although the later data are of poorer quality .
we assembled three representative average spectra .
for the early outburst , we combined spectra from 1991 january 1724 ; no obvious spectral differences are visible between these .
we produced a second epoch spectrum from february 628 .
this corresponds to some way down the decline , but before the secondary maximum in the x - ray lightcurve .
for the third sample we use a single epoch , march 2728 , within the secondary maximum , which was seen in the uv data @xcite . at later epochs , the _ iue
_ data are of poor quality , and the _ hst_spectrum ( see next section ) is of more use .
we did not attempt to re - derive a reddening for gs 1124684 from the _ iue _ data as the _ hst_spectrum of this source is of much higher quality . for gro j0422 + 32 ,
the coverage was less complete ; the first two months were observed regularly , but there was only one more observation about a year later .
we assembled two average early spectra from 1992 august 1626 and 1992 september 227 respectively .
the last visit lies beyond the main exponential decay phase of the outburst .
since the reddening derived by @xcite from the 2175 feature was inconsistent with other estimates , we combined these two spectra and reanalyzed them to attempt to resolve the discrepancy . @xcite
derived @xmath47 , but did so by assuming the entire _ iue _ spectrum was a power - law .
consequently there is a danger that the derived reddening may be biased by a failure of the model . as for xte j1859 + 226@xcite , we tried using a range of different wavelength ranges and both a power - law and irradiated disk model in order to test for such sensitivities .
we do indeed find that a wide range of reddening values are derived , 0.290.48 , dependent mainly on the choice of fitting range .
as can be seen from the fit by @xcite , a power - law model does fail both at short and long wavelengths .
we find that using the 16503120 range adopted for xte j1859 + 226 , significant systematic residuals are seen at the longer wavelengths
. fits with this region ( @xmath85 ) included produce large values of @xmath86 , whereas if this region is excluded a range of 0.290.38 is found .
the range above 2500 contains little if any information on the 2175feature , and the discrepancy with other estimates is also largely resolved if it is excluded , so we choose to use fits without this region .
_ hst _ spectra from the main part of the outburst are available for two of our sources : gs 1124684 @xcite and xte j1859 + 226@xcite .
the data on gs 1124684 were obtained with the faint object spectrograph ( fos ) fairly late in the outburst , but still in the exponential decay phase .
we obtained these data from the st - ecf archive .
they were recalibrated by the archive using the poa_calfos pipeline @xcite and the final reference files . as for other sources
, we reanalyzed the 2175 feature .
@xcite found a good fit using a power - law model and we find the same ; the irradiated disk model which worked better for xte j1859 + 226 provides a very poor fit for gs 1124684 .
a reddened power - law fit to the 15743500 range ( chosen to exclude any weak balmer jump contribution ) gave @xmath87 , @xmath88 .
the reddening estimate is not significantly different to that of @xcite , but the derived spectrum is somewhat bluer .
this is due to the different extinction curve used ; we use that of @xcite , whereas @xcite used the @xcite curve . the difference between these curves
is shown by @xcite , and the different spectral slopes derived is representative of the real uncertainty in the process ( see section [ dereddenvarsection ] for more discussion of this point . )
we adopt this value , with 20percent uncertainty , as a reddening measure for gs 1124684 : @xmath42 .
we also repeated the fit with the full dataset , 15744816 , and with all the wavelength ranges used for xte j1859 + 226 . in every case , we derive @xmath5 consistent with the range @xmath42 .
a full analysis of data obtained with the space telescope imaging spectrograph ( stis ) of xte j1859 + 226 along the lines here has already been presented by @xcite , we simply include selected results of that work for comparison with the other sources presented here .
these spectra span the full 115010300 range at several epochs through the outburst .
these are the only data on this source included .
llrllcl object & date & outburst & facility & wavelength & plotted & reference + & & day & & range used ( ) & & + a 062000 & 1975 sep 24 & 2022 & palomar , hale 5 m & 320010350 & y & 1 + & 1975 dec 811 & 117120 & palomar , hale 5 m & 320010350 & y & 1 + gs 1124684 & 1991 jan 15 & 0 & eso , ntt & 43507500 & y & 2 + & 1991 jan 18 & 3 & eso , 1.5 m & 38008600 & a & 3 + & 1991 jan 19 & 4 & eso , 1.5 m & 38007400 & a & 3 + & 1991 jan 20 & 5 & eso , 1.5 m & 63008600 & a & 3 + & 1991 jan 23 & 8 & eso , ntt & 42006200 & a & 3 + & 1991 feb 15 & 31 & eso , 1.5 m & 38507450 & y & 3 + & 1991 feb 21 & 37 & eso , 1.5 m & 38509850 & n & 3 + & 1991 mar 6 & 50 & eso , ntt & 40008150 & y & 3 + & 1991 mar 21 & 65 & eso , 2.2 m & 40509050 & n & 3 + & 1991 mar 25 & 69 & eso , ntt & 61008100 & n & 3 + & 1991 may 5 & 110 & eso , 1.5 m & 40007100 & n & 3 + & 1991 may 19 & 124 & eso , 3.6 m & 42009850 & y & 3 + gro j0422 + 32 & 1992 aug 17 & 3 & lowell perkins , 1.8 m & 38008500 & y & 4 + & 1992 aug 18 & 4 & lowell perkins , 1.8 m & 48506800 & n & 4 + & 1992 sep 25 & 42 & lowell perkins , 1.8 m & 48506800 & n & 4 + & 1992 oct 10 & 57 & lowell perkins , 1.8 m & 48506800 & y & 4 + & 1992 nov 4 & 82 & lowell perkins , 1.8 m & 48506800 & n & 4 + & 1992 dec 15 & 123 & lowell perkins , 1.8 m & 48506800 & y & 4 + & 1993 jan 21 & 160 & lowell perkins , 1.8 m & 48506800 & n & 4 + & 1993 feb 11 & 181 & lowell perkins , 1.8 m & 48506800 & y & 4 + grs 100945 & 1993 nov 18 & 64 & eso , 2.2 m & 36009050 & y & 5 + grs 1716249 & 1993 oct 58 & 58 & eso , 1.5 m & 2.2 m 2 & 46006700 & y & 6 + the highest quality calibrated spectra for a 062000 were presented by @xcite who show spectrophotometry from several epochs , of which two , around 21 and 118 days after outburst peak respectively , were obtained during the exponential decay phase of the outburst .
the original data are no longer available , so these two spectra were digitized from the paper and included in our sample .
the first observation occurred close to the ans uv observations .
@xcite present two low - resolution seds of the optical .
these are of poorer quality than those of @xcite , and are only shown with an arbitrary flux scale , so we do not include these .
a large sample of optical spectra were presented by @xcite and @xcite .
these span days 0124 days after outburst peak .
the spectrum included in @xcite was digitized ; those from the later paper were kindly supplied by the authors . for the latter ,
the reddening correction pre - applied was removed before dereddening them as described above for consistency . for further analysis , we merged the spectra from 1991 jan 1823 .
we also rejected the 1991 mar 25 spectrum which has limited wavelength coverage , and a very different slope to all others considered , even 1991 mar 21 .
some portions at the extreme near - ir and near - uv ends of some spectra show abrupt and non - repeatable transitions and these were also excluded .
@xcite present spectra from 1992 august 16 to september 25 .
we include an absolutely fluxed spectrum from 1992 august 17 , from this dataset , which was provided by the author .
the remaining epochs were digitized from a figure also provided by the author as the original data were not available .
grs 100945 suffered both from a relatively short exponential phase of outburst , @xmath89days @xcite , and from a late discovery .
the only spectrum from the exponential decay phase was presented by @xcite , spanning 35009500 .
this spectrum was digitized , placed on an absolute flux scale ( assuming 1 unit @xmath90 @xmath91erg@xmath44s@xmath92 ; della valle , private communication ) , and corrected to our preferred dereddening solution .
one flux calibrated spectrum from the exponential decay phase was presented by @xcite .
this spectrum was digitized and corrected to our preferred reddening value and extinction curve as for other sources .
our combined sample of optical spectra are shown in fig .
[ optsedfig ] . for these purposes ,
satellite spectra have been truncated at @xmath93hz , the approximate highest value recorded for a 062000 from the ground .
we have excluded some spectra for which the flux calibration appears incorrect ; when there are large distortions which are not present at any other epoch ; and others to avoid duplication of two very similar spectra .
it is clear that after applying the source selection criteria described above , and subject to these additional restrictions , the spectra define a relatively homogeneous sample .
all are blue , with @xmath94 increasing toward the uv .
all have an approximately power - law form .
subject to the large uncertainties in distance described above , all represent comparable optical luminosities . to perform a more quantitative comparison , we must parameterize and fit the spectra . at this stage
we consider only very simple characterizations ; we will calculate model disk spectral energy distributions in section [ modelsection ] , but uv coverage is required to adequately constrain such models .
the two simple characterizations widely used in the literature are a power - law or a hot black body ; the latter has an approximately power - law form on the rayleigh - jeans tail , and so can also provide an adequate fit to the data .
given that a single temperature black body is likely a very poor description of an accretion disk , the black body fit has limited physical significance , but is useful for comparison with other work . the temperatures that we derive are somewhat lower than other authors have obtained in this way . in the case of
a 062000 we estimate 1300015000k , compared to 2500030000k obtained by @xcite . for both grs 1716249 and
100945 we estimate a temperature of @xmath95k , compared to published estimates of 20000k @xcite and 25000k @xcite respectively .
the differences likely reflect the difficulty of reliably constraining such high temperatures using optical data alone , and in this regime large derived temperature differences can result from quite small differences in the data and its treatment ( for example the wavelength range used in fitting . ) in view of this difficulty , and the limited physical meaning of a single temperature fit , we ignore the results of these fits from hereon ; suffice to say that _ the spectra indicate a hot continuum , sufficiently hot that optical observations have very limited discriminatory power alone . _
lcccc[t ] object & outburst & & black body + & day & @xmath29 & @xmath96 & t ( k ) + a 062000 & 2022 & @xmath97 & @xmath98 & 14700@xmath99 + & 117120 & @xmath100 & @xmath98 & 12900@xmath99 + gs 1124684 & 0 & @xmath101 & @xmath102 & 14100@xmath103 + & 38 & @xmath104 & @xmath105 & 17100@xmath99 + & 31 & @xmath106 & @xmath102 & 17000@xmath107 + & 37 & @xmath108 & @xmath109 & 15300@xmath110 + & 50 & @xmath111 & @xmath102 & 19800@xmath103 + & 65 & @xmath112 & @xmath105 & 11800@xmath113 + & 110 & @xmath114 & @xmath109 & 31500@xmath115 + & 124 & @xmath116 & @xmath109 & 20100@xmath117 + gro j0422 + 32 & 4 & @xmath118 & @xmath119 & 17700@xmath117 + & 5 & @xmath120 & @xmath121 & 18300@xmath103 + & 42 & @xmath122 & @xmath121 & 13100@xmath113 + & 57 & @xmath123 & @xmath121 & 13100@xmath99 + & 82 & @xmath124 & @xmath121 & 13100@xmath107 + & 123 & @xmath125 & @xmath121 & 12600@xmath113 + & 160 & @xmath126 & @xmath121 & 11800@xmath113 + & 181 & @xmath127 & @xmath121 & 12400@xmath128 + grs 100945 & 64 & @xmath129 & @xmath130 & 17000@xmath131 + grs 1716249 & 5 & @xmath132 & @xmath133 & 17200@xmath134 + xte j1859 + 226 & 2 & @xmath135 & @xmath136 & 28000@xmath110 + & 11 & @xmath137 & @xmath136 & 22500@xmath107 + & 21 & @xmath138 & @xmath136 & 19900@xmath107 + & 115 & @xmath139 & @xmath140 & 17400@xmath110 + & 141 & @xmath141 & @xmath140 & 17100@xmath134 + power - law fits have previously been performed for some of our sources .
they provide a more empirical description of the data , but one which more directly relates to its information content
. for gs 1124684@xcite found an initially blue spectrum , @xmath142 ( where @xmath143 ) , which softened to @xmath144 after the secondary maximum .
@xcite also remarked that this object exhibited a period at the beginning of the outburst when the source had a steeper ( bluer ) spectrum than later observed . for
gro j0422 + 32 , @xcite find the optical spectral index softened from 0.6 to 0.0 over the first three months of the outburst , then stabilized .
@xcite found spectral indices of between -0.1 and -0.4 for most of the outburst , although the evolution is less clear in their data and they assumed a relatively low extinction , @xmath145 .
we fit power - laws to our optical spectra and tabulate power - law indices in table [ optfittable ] .
these values differ somewhat from those quoted above , as absolute values are rather sensitive to the assumed @xmath5 .
some spectral indices have also been plotted as a function of time in fig .
[ plevolutionfig ] .
all of the values that we derive fall within the approximate range @xmath146 , and it should be noted that _ all _ are steeper ( bluer ) than the canonical @xmath147 of a steady - state viscously heated disk which is sometimes assumed
. some variations within this range may be due to incorrect dereddening , but the full range can not be explained in this way , and the variations between repeated observations of a single source indicate that there are real differences .
there is no single ` correct ' value , and so the utility of assuming a standard spectral slope ( or photometric color ) in estimating reddening values is limited .
the observed variation corresponds to an intrinsic @xmath33 color of 0.00 to -0.24 .
this range is small enough that it does have some value for an approximate reddening estimate for sources with relatively large reddening , but one should still be cautious , as we have only obtained this degree of uniformity by _ a posteriori_selection of sources by outburst behavior and orbital period .
for example , the unusual source xte j1118 + 480 has a somewhat flatter optical spectrum @xcite and the long - period system gro j165540 is much redder in the optical @xcite ) .
the dashed line corresponds to the canonical @xmath2 value for a steady - state viscously heated disk ; all spectra appear steeper than this .
the dotted line is shows an exponential decay fitting the data for xte j1859 + 226 , and is intended only to guide the eye in comparing these data with those for gro j0422 + 32 , for which the line has been offset downward by 0.45 . ] for a 062000 there are too few points to draw any convincing conclusions about evolution of the optical spectrum , other than that little change appears to have occurred between the two epochs .
this conclusion is borne out on longer timescales by the photometry of @xcite who find approximately constant @xmath21 values around @xmath148 throughout the outburst , corresponding to a dereddened color of @xmath149 , comparable to our spectral shapes . for gs 1124684there are more data , but these exhibit rather large scatter .
the spectra may be hardening , but it is hard to say with confidence , and the photometric analyses of @xcite and @xcite suggest the opposite behavior .
gro j0422 + 32 and xte j1859 + 226 show more coherent trends , but this may reflect better sampling and/or data quality rather than an intrinsic difference .
the latter in particular , being based solely on satellite data , should be robust .
both sources clearly become softer optically during the outburst .
there appears to be rapid evolution during the first month , then slower softening ( at least for gro j0422 + 32 ) later .
this evolution has already been noted by @xcite , although all of their power - law indices are softer than ours due to a lower assumed reddening .
the photometry of @xcite may also indicate an initial rapid softening .
this behavior suggests that the optically bright regions are cooling during the outburst .
we will defer more detailed discussion until we consider the combined optical and uv spectra of these sources ( section [ modelsection ] ) . although the evolution is very similar for gro j0422 + 32 and xte j1859 + 226 , the actual values are rather different .
this difference may be real , or may reflect an incorrect dereddening of the spectra .
the dotted lines plotted in fig .
[ plevolutionfig ] are offset by 0.45 . at late times
gro j0422 + 32 is about a further 0.25 below the plotted line .
uncertainty in reddening introduces an uncertainty in the spectral slope of @xmath150 for both sources , however , so most of the difference in the optical slopes could be due to an incorrect @xmath5 .
we will reexamine the difference in the light of the uv spectra and models in section [ modelsection ] .
we have invested some time in comparing the optical results in isolation as these are often all that is available .
in addition these results can be compared directly with optical photometry which typically has better temporal sampling .
based on optical results alone , the objects appear relatively homogeneous and we can not be confident that apparent differences are not due to incorrect dereddening .
optical colors do give some useful information about spectral _ evolution _ independent of reddening uncertainty .
they do not allow us to distinguish between different models robustly however ; for that we must extend the data into the uv .
based on the analysis so far , gro j0422 + 32 and xte j1859 + 226 exhibit definite and similar softening during the outburst , whereas a 062000 and gs 1124684 exhibit no clear trends , although very little data are available for the former , and the latter exhibit large scatter .
our combined sample of broad - band spectra is shown in fig .
[ uvsedfig ] . we have selected data for which both optical and uv coverage is available .
where possible we have used contemporaneous data , but in some cases , particularly for a 062000 , the time difference is significant .
it can be seen that the power - law fits to the optical data do not adequately extrapolate for any object .
all uv spectra flatten with respect to the optical .
the degree of flattening varies , with a 062000 and gs 1124684 continuing to rise more slowly in the uv , while gro j0422 + 32 and xte j1859 + 226 actually turn over and begin to fall .
one might question whether this difference is real or reflects incorrect dereddening , but since we see both forms in xte j1859 + 226 , it is likely a real variation within the sample .
we can characterize this distinction by dividing the spectra into two optical / uv states .
we refer to spectra in which the far - uv spectrum continues to rise as uv - hard , and ones in which it turns over and declines as uv - soft .
the presence of such a turnover is natural and is a consequence of the cutoff in the disk temperature distribution at the disk edge ( or at a cooling front ) .
we will discuss the interpretation of these spectra in section [ modelsection ] .
first , however , we examine more carefully how the spectra are affected by uncertain dereddening .
the effect of uncertain dereddening is easiest to quantify in the optical , as optical extinction curves are less diverse than those in the uv , and the variation of extinction with wavelength is approximately monotonic .
uncertainty in @xmath5 will introduce an uncertainty in the optical spectral slope , whereas an uncertain @xmath28 will affect the implied luminosity level . note that these effects are semi - independent since @xmath151 may not always have the galactic average value of 3.1 . to quantify the effect on the spectral slope , for each source we have shown in fig .
[ optsedfig ] a bow - tie indicating the power - law slopes derived if @xmath5 is allowed to vary within the uncertainties estimated .
we have also included a corresponding estimate of the change in power - law index , @xmath96 , in table [ optfittable ] .
[ [ uv ] ] uv ~~ the effect on the uv spectrum is more complex , as the uv extinction curve exhibits several features , and the relative strength of these can vary .
the effect of uncertain @xmath5 , assuming a fixed extinction curve , can be quantified in the same way as for the optical data , although the form is not simply a variation in power - law slope .
this effect is shown by the curved bow - ties in fig .
[ uvsedfig ] .
note that there is no increase in the uncertainty around the 2175bump as this region is explicitly flattened out in dereddening the spectra ; a larger @xmath5 would only be consistent with the data if the bump were unusually weak and vice versa .
@xcite has presented a convenient parameterization of the variations in the extinction curve , and also presents a recipe for quantifying the effect of the resultant uncertainty in the derived sed .
@xcite have already applied this analysis for xte j1859 + 226 .
we show comparable results for gs 1124684 and gro j0422 + 32 in fig .
[ uvextcurvefig ] .
note the uncertainty in the extinction curve is less of a problem for a 062000 for which a line - of - sight curve was determined in parallel with the reddening determination @xcite .
we have shown that while a single power - law is an acceptable description of most optical spectra , it does not adequately describe the uv break , which seems to be ubiquitous .
this break is to be expected .
we believe that the spectra included in this sample should reflect the spectrum of an accretion disk in a hot state .
the hot region should terminate at effective temperatures between 6000k and 10000k , as regions at this temperature should be unstable . if such a cut - off occurs , then the optical spectrum will lie on the rayleigh - jeans tail for most of the disk area hence exhibits a steeper ( bluer ) spectrum than at higher frequencies , as observed . we can attempt to quantify the disk temperature distribution using the commonly used model for a disk heated by both viscosity and irradiation and emitting locally as a black - body .
this model , its origins , and original references , are summarized by @xcite , and we will not repeat the discussion .
an alternative discussion is provided by @xcite .
we use that model exactly as described by @xcite , and with the same definitions and terminology .
we emphasize that the model actually only describes a disk with two forms of heating , one of which drops off as @xmath152 ( viscous heating ) and one as approximately @xmath153 ( irradiation heating ) .
the exact model used ( based on @xcite ) assumes a slightly slower drop - off of the latter and is adopted for consistency with earlier work and as a simple empirical description , rather than as a theoretically motivated formulation ( see @xcite ) . in practice
the exact prescription adopted has little effect on the optical and uv spectra , and so any error introduced in this way will be less than that due to uncertainties in dereddening the spectra .
the essential discriminating evidence between these two effects is the shape of the far - uv spectrum ; a uv - hard spectrum which continues to rise favors @xmath152 heating , whereas a uv - soft spectrum with a clear peak indicates @xmath153 heating . beyond the peak
, the uv - soft spectrum will tend towards a @xmath154 spectrum or steeper , but this form will not be realized within the observable spectral range .
the @xmath152 component , which gives rise to the classical @xmath2 accretion disk spectrum @xcite , is usually assumed to be due to viscous heating , but could equally describe irradiation which drops off in a similar way with radius .
indeed this functional form is commonly adopted in modeling irradiated disks in protostars ( e.g. @xcite ) .
we will return to this alternative interpretation in section [ coronasection ] .
the analysis we will perform is virtually identical to that done by @xcite for xte j1859 + 226 and we will directly compare to those results without reanalyzing those data . one important difference , however , is that xte j1859 + 226 is the only source for which a full uv optical spectrum was obtained quasi - simultaneously with _
hst_. for the other objects in our sample having uv coverage , the optical spectra were taken from the ground days or even weeks apart .
the non - simultaneity compromises the reliability of the seds obtained , and we will discuss the severity of this problem on a case - by - case basis . in general it requires that we introduce an additional free parameter representing the difference in flux levels between the two epochs , but it should be remembered that spectral changes might have occurred as well .
it is worth emphasizing that we explicitly assume there are no other sources of flux besides the accretion disk ; emission from a jet or heating of the companion star are neglected . for the sources and wavelength region adopted ,
this is a reasonable assumption , and was a major criteria in their selection .
llrcrclc object & dates & & & + a 062000 & 2048 & 0 & 06000 & 7400 & 70007800 & 11.35 & 9.6513.35 + gs 1124684 & 29 & 0 & 06000 & 10000 & 860011400 & 0.85 & 0.611.26 + gs 1124684 & 2244 & 0 & 05000 & 7800 & 70008600 & 0.98 & 0.751.31 + gs 1124684 & 120 - 144 & 7000 & 50009000 & 8400 & 80008800 & 0.23 & 0.190.27 + gro
j0422 + 32 & 210 & 12600 & 1140014200 & 5200 & 30005600 & 1.31 & 0.981.72 + gro
j0422 + 32 & 1944 & 9400 & 840011000 & 5000 & 46005600 & 1.98 & 1.282.78 + a 062000 has a high quality optical sed , and a well determined uv extinction , so in principle it should be one of the best suited objects to our study .
it suffers , however , from very crude wavelength coverage in the uv and , most severely , from the lack of simultaneous uv and optical observations .
we will compare optical data taken around day 21 of the outburst with uv data from around day 47 .
the 26 day interval between them may mean that the spectrum has changed significantly .
a particular concern is that the uv observation happened around the time of the onset of a secondary maximum in the x - ray lightcurve ( e.g. @xcite ) .
the optical spectrum taken later , around day 118 , is very similar to the early one however , suggesting that little evolution actually occurred in the optical spectrum , and that the comparison is valid . keeping these concerns in mind
, we proceed to fit a disk model to the sed , allowing for an arbitrary offset between the optical and uv data .
this approach is equivalent to assuming that the only difference between the two epochs was in the flux level of the spectrum , rather than the shape .
the fits are shown in fig .
[ nmonfitfig ] .
the best fit model has no irradiative heating and corresponds to the classic viscously heated steady - state disk . for comparison
we also show a purely irradiated model .
this is an extremely poor fit , even with an arbitrary offset between the optical and uv . improving the uv fit would require an increase in the irradiation temperature and would steepen the optical spectrum .
of course some optical spectral evolution may have occurred .
the later spectrum is flatter , however , and other sources also tend to flatten as the outburst proceeds .
flattening of the optical spectrum would further degrade the irradiated fit , so it seems unlikely that the irradiated model can be accommodated in this way .
, i.e. a purely irradiated disk .
the latter clearly can not fit well . to improve
the uv fit would require a large increase in the irradiation temperature and would then steepen the optical spectrum to an unacceptable degree . ] the implied parameters ( for the uv epoch ) are sensible .
the outer edge viscous temperature , @xmath155k , is comparable to that found in xte j1859 + 226 late in outburst , and is close to the expected minimum temperature of the hot phase of the disk @xcite .
the normalization can be interpreted as described by @xcite . assuming a distance @xmath156kpc , and inclination @xmath157 @xcite
, we estimate a disk radius @xmath158 cm , comparable to the expected tidal truncation radius of the disk ( @xmath159 cm ) .
the implied disk radius would increase to @xmath160 cm when the model is normalized to fit the optical data .
we can also estimate the mass flow rate required to provide this optical luminosity as described by @xcite . at the uv epoch
, we estimate @xmath161m@xmath162yr@xmath92 . assuming an 11.0m@xmath162 black hole @xcite , and 10%accretion efficiency
, the eddington limit corresponds to @xmath163m@xmath162yr@xmath92 , so we obtain a mass flow rate a few tenths of the eddington limit . given the uncertainties , the observations are thus consistent with a lobe - filling , viscously heated disk .
contrary to common assertions , there seems to be no need for irradiative heating to provide the energy radiated in the optical - uv .
this is not to say that irradiation plays no role ; it may still be crucial in stabilizing the disk against a transition to a cool state .
the point is that the main source of energy emitted is consistent with viscous heating rather than irradiation .
finally , we note that the long wavelength spectrum is flatter than the model predicts .
this difference is relatively subtle , and may not be of great significance .
however , a similar long - wavelength flattening is seen in xte j1859 + 226 @xcite and may indicate a weak contribution from synchrotron emission @xcite .
gs 1124684 has better temporal coverage than a 062000 , and generally optical observations can be selected from within the time - period spanned by multiple _
iue _ observations .
in addition , the _ hst_spectrum obtained later in the outburst is of higher quality , albeit lacking the shortest wavelengths .
two concerns are manifest here , however .
firstly , we have already seen that the power - law slopes derived from the optical spectra scatter significantly .
the most likely interpretation of this scatter would be that the flux calibration is not totally reliable .
if this is correct then the spectral fits are somewhat compromised .
secondly there is a sharp upturn in the far - uv , rather similar to that seen in xte j1859 + 226when the latter was corrected with the pure @xcite extinction curve .
@xcite interpreted the upturn in that case as due to a far - uv rise component of extinction which is weaker than in the @xcite extinction curve .
the same could be true for gs 1124684 , and similarly introduces uncertainty in the dereddened far - uv spectrum .
we therefore omitted these points from the fits .
, i.e. a purely irradiated disk . ]
disk model fits are shown in fig .
[ nmusfitfig ] , performed as for a 062000 .
gs 1124684 is also best fitted by a model dominated by viscous heating , exhibiting uv - hard spectra at all epochs . a model with purely irradiative heating would not fit .
the disk outer edge temperature is comparable to that in a 062000 at late times , but higher earlier , dropping from @xmath164k to @xmath165k ; a similar decline was found by @xcite .
assuming @xmath166kpc and @xmath167 ( @xcite ; @xcite ) we obtain disk radii of 1.43.0@xmath168 cm , comparable to the expected tidal truncation radius of @xmath159 cm . the scatter inferred in the radius may arise from the uncertainty in the optical spectral slope , since this uncertainty will affect the derived temperature , which in turn will be inversely correlated with the normalization .
these data thus also appear approximately consistent with a lobe filling disk .
the higher temperatures early on may indicate that the disk is indeed being tidally truncated , since the hot - phase of the disk should be able to extend to below 10000k if it were unbounded .
the temperatures we estimate are higher than for a 062000 , even though the disk radius is expected to be comparable .
a higher temperature implies a larger mass flow rate ( if the spectrum is indeed due to viscous heating ) .
note that this conclusion is largely independent of distance uncertainties , provided both disks are tidally truncated .
@xcite indirectly provide support for this conclusion .
they compare the x - ray lightcurves and estimate a flux scaling factor of @xmath169 for the ratio of a 062000 to gs 1124684 .
they argue that the overall flux level of the two outbursts should be the same , assuming that the eddington scaled luminosity of the intermediate state should be constant .
this is only expected , however , for extreme parameters , for example a black hole mass in a 062000 of just 4.5m@xmath162 is assumed . if we instead adopt current best estimates of the parameters @xcite , then we instead predict a flux scaling of @xmath170 , if the eddington scaled luminosities were the same
. this scaling is uncertain , but significantly larger than observed , implying that a 062000 is underluminous relative to gs 1124684 , and hence that the accretion rate in gs 1124684 was likely larger .
@xcite also note that the ratio of optical to x - ray flux is a factor of 1.6 larger in a 062000than in gs 1124684 , and attribute this difference to more efficient irradiation of the outer disk in a 062000 .
there is another explanation , however .
both the optical and x - ray bandpasses lie at the extreme ends of the spectral energy distribution , on the rayleigh - jeans and wien tails respectively .
if the mass flow rate is higher , the effect is to increase the temperature throughout the disk and shift the sed out of the optical bandpass and into the x - ray one .
such a difference is supported by the hotter optical / uv spectra inferred in gs 1124684 .
this spectral shift will naturally decrease the optical to x - ray flux ratio without invoking any other changes .
for example , consider two viscously heated disks around 10m@xmath162 black holes , with inner and outer radii of 3r@xmath171 and @xmath172 cm respectively .
we adjust the mass flow rate to give outer disk temperatures of 8000k and 12000k respectively and calculate the ratio of @xmath173 band to x - ray flux ( 1.37.5kev ) .
the ratio can change by at least a factor of two and possibly more , depending on the relative weighting across the x - ray bandpass and the effect of spectral hardening on the x - ray spectrum .
this effect is more than adequate to explain the difference in the flux ratios between a 062000 and gs 1124684 without any irradiation effect .
if we assume a 6.95m@xmath162 black hole in gs 1124684 , then we estimate mass flow rates of @xmath174m@xmath162yr@xmath92 , @xmath175m@xmath162yr@xmath92 , and @xmath176m@xmath162yr@xmath92 for epochs 13 respectively ( comparable to that obtained by @xcite assuming similar parameters ) .
these values are superficially somewhat problematic , as the eddington limit should be reached for @xmath177m@xmath162yr@xmath92 , assuming 10%accretion efficiency . even with higher efficiencies , the early phase of the outburst
should thus have involved super - eddington accretion .
this is not a fatal problem , however , since the eddington limit strictly only applies to spherical accretion and a geometrically thin disk could deliver mildly super - eddington accretion rates . such a super - eddington model was indeed considered for gs 1124684 by @xcite , and @xcite also assumed a peak accretion rate somewhat above the eddington limit .
a more pertinent question is whether such a large accretion rate can be reconciled with the observed thermal ( disk ) component in the x - ray spectrum of gs 1124684 @xcite .
@xcite did address this question in the context of their spectral modeling and show that the two are consistent provided that compton scattering modifies the x - ray spectrum with a spectral hardening factor @xcite of @xmath178 .
this factor is larger than the value of 1.7 usually adopted , but @xcite argued that values as large as 3 are plausible . if one accepts this correction then both the optical / uv spectrum , and the soft x - ray thermal component can be consistently interpreted with a purely viscously heated disk . as noted above for a 062000 , irradiation may still play a role in stabilizing the disk in the high viscosity state , but it is not required to provide the main energy input .
we will consider an alternative explanation for the optical / uv spectrum in section [ coronasection ] . for gro j0422 + 32
we have two useful epochs with both _
iue _ and optical data .
these data may also suffer from similar problems with the far - uv rise component of extinction to gs 1124684 and gro j0422 + 32 , but in the opposite sense , as there seems to be an enhanced drop off in flux at the shortest wavelengths
. fits to these data are shown in fig .
[ nperfitfig ] .
strong irradiative heating is needed in this case , although as the outburst declines viscous heating may be becoming important as in xte j1859 + 226 ; the timebase and data quality are inadequate to say this with any confidence , however .
the irradiation temperatures inferred are lower than for xte j1859 + 226 , with @xmath179k early in outburst , compared to @xmath180k in xte j1859 + 226@xcite .
lower temperatures are consistent with the softer optical spectrum earlier estimated . as seen in fig .
[ uvsedfig ] , this difference does seem to be reflected in the lower peak frequency of the sed , so is likely real . as for the other sources
, we can estimate the disk radius required ; assuming @xmath181kpc and @xmath182 @xcite , we derived disk radii of @xmath183 cm and @xmath184 cm for the first and second epochs respectively .
these numbers seem too large , as we estimate a tidal truncation radius of @xmath185 cm assuming the parameters of @xcite .
this discrepancy could be resolved with a reduction in the source distance ( which is not totally impossible ; @xcite ) , with a more modest increase in the temperature ( which is factored into the large uncertainty in this derived radius ) , or if the inclination uncertainty is larger than assumed .
we have seen that uv - hard spectra are ubiquitous , and can occur even when a bhxrt is at a high optical luminosity .
it is common , at least in the field of x - ray binaries , to interpret such a spectrum as a signature of a viscously heated steady - state disk ( e.g.@xcite ) , and thus far we have done the same .
this is not implausible , particularly in the case of a 062000 for which the required mass flow rate is not large .
all this spectral form really indicates , however , is the temperature distribution , and that the heating rate varies as approximately @xmath186 .
an alternative irradiation model is used by the young star community which actually predicts irradiative heating varying as @xmath186 ( e.g. @xcite ) .
this model addresses the case where a circumstellar disk is irradiated by the stellar surface .
the key difference is that since the star is not compact , it is not a point source , but an extended one which illuminates the disk from above . indeed a similar case which also produces @xmath186 heating arises from a ` lamppost ' model , where the source of x - rays are raised above the disk ( e.g. @xcite ) . as described in the appendix
, one can produce uv - hard spectra with an elevated ( or vertically extended ) x - ray source .
this behavior is generically expected , except for the specific case of a concave disk with a sufficiently compact x - ray source , possibly mediated by local scattering by an extended disk atmosphere .
it is worth emphasizing that where irradiation is attributed to x - rays scattered by a central coronal source , one _ expects _
@xmath186 heating , in contrast to the assumption usually made in modeling outburst behavior that irradiation drops off approximately as @xmath187 ( e.g. @xcite , @xcite , @xcite , @xcite ) .
if x - ray scattering occurs in a highly vertically extended disk atmosphere , however , @xmath187 heating could still be appropriate .
how plausible are these scenarios ?
x - ray scattering ` coronae ' are certainly believed to be present in x - ray binaries .
however , it is likely that coronae are not spherical , but very flattened , and represent a vertical extension of the disk atmosphere ( e.g.@xcite ) .
effective illumination of a convex disk , with a vertically extended central corona , thus seems difficult to achieve , and hence this seems unlikely as a mechanism for producing uv - hard spectra .
finally , we note that x - rays from a jet might provide an alternative ` lamppost ' illumination , either by direct synchrotron emission or by scattering .
relativistic beaming might then be expected to considerably reduce the back - illumination of the disk , but large luminosities are not required , especially as the x - ray source could then be higher above the disk than is plausible for a corona .
@xcite suggested that warping of disks may facilitate irradiation of disks which would otherwise be self - shielded , and this could also be expected to modify the radial dependence of the heating . @xcite
have examined this question quantitatively .
they do find strong variations in the radially averaged heating , but these are not as simple as the additional @xmath188 dependence that would be required to reproduce the @xmath189 spectra , although they do find that an average reprocessing efficiency of @xmath190 can readily be produced with a warped disk .
while only a fraction of the disk is illuminated at any time , the effect of irradiation on the thermal structure of the disk will last for of order the local thermal timescale .
this should be comparable to the local dynamical timescale ( i.e. the keplerian rotation period ) , so heating effects will be azimuthally averaged as material rotates through the illuminated patch and hence the whole disk will be affected ( g. dubus , priv .
. further investigation of the spectral energy distributions expected in the warped disk case is needed .
@xcite do note , however , that the short - period bhxrts we are considering here are expected to be stable against radiation driven warping , so in the absence of another warping mechanism , it remains questionable whether warping can be an important factor for these objects .
in light of the above discussion , we should reconsider the origin of the optical emission in bhxrts in outburst .
it is sometimes stated that optical emission is always totally dominated by reprocessed x - rays .
this may be true in neutron star systems , but there is good reason to _ expect _ irradiation to be weaker in the black hole case , as there is no direct emission from the compact object surface @xcite .
this difference is invoked by these authors to explain the observations that black hole lmxbs are transient , whereas most analogous neutron star systems are persistent x - ray sources .
the studies presented here suggest that the situation may be more marginal than usually assumed .
if one interprets a @xmath2 spectrum as a signature of a viscously heated disk , then the answer is clearly no , as this spectrum occurs rather often among the population . in the case of a 062000 , at least
, if one takes a plausible mass flow rate , and plausible disk size then viscous heating alone can readily explain the observed optical brightness .
in the case of xte j1859 + 226 , the sed can initially be interpreted as irradiation dominated , but a transition to the viscously dominated regime appears to occur suggesting that the irradiation dominance is rather marginal .
of course this conclusion hinges upon one interpretation of the sed , but the alternatives considered are relatively unpalatable .
irradiation by x - rays from a lamppost ( perhaps the jet ? ) , or by a vertically extended corona could reproduce the observed sed , but requires an excessive luminosity for the extended source and/or a rather large vertical extent .
we thus appear left with viscous heating of the outer disk as the most natural explanation of the observed optical flux during outburst , at least when uv - hard spectra are seen .
as noted several times , irradiation should still play a role in maintaining the disk in a hot , high viscosity state by suppressing the disk instability , but it does not appear to dominate the energy budget . the point is often made that studies such as this one are severely compromised by assumption of local black - body spectra . at some level , of course , this is true , but in the uv , and especially the optical , the effect is likely to be mild .
the two most important effects on a optical - uv broad - band sed , for the temperatures considered , are hydrogen breaks and limb - darkening .
line - blanketing by many blended absorption lines could have some effect , although the highest quality data suggest a rather clean continuum where the effect of intrinsic ( non - interstellar ) absorption lines is very weak or absent , except for ly@xmath29 @xcite .
the balmer jump falls well within our spectral coverage .
while this results in strong distortion of the sed of hot stars , such large effects are clearly not present in our data , and there is little if any evidence for its presence at all .
the observations are consistent with spectral models of accretion disks around stellar mass black holes which do predict rather weak balmer jumps ( e.g. @xcite ; hubeny priv . comm . )
a much larger effect is predicted at the lyman break , and has the effect of significantly enhancing the far - uv flux longward of the break .
this effect produces an upturn which is principle is detectable .
such an upturn was indeed observed in xte j1859 + 226 @xcite , and may also be present in gs 1124684 , as shown earlier .
an alternative explanation was advanced by @xcite , however , that this upturn originates from deviation of the far - uv rise component of extinction from the galactic average .
disentangling the two effects is likely impossible without line - of - sight extinction curves or a larger sample of objects .
furthermore , strong lyman breaks can also be predicted in agn spectra , but are not observed ( see @xcite and references therein ) .
the other deviation from black body spectra that could be significant is limb - darkening @xcite .
this is wavelength dependent and stronger in the uv , so tends to soften the far - uv flux .
it is an inclination dependent effect , however , and for the moderate inclinations of the objects considered here ( 4060@xmath191 ; see section [ modelsection ] ) should not be dominant . with both balmer and lyman breaks and limb - darkening
, the models of @xcite do not deviate from their black - body counterparts by more than @xmath192 in @xmath193 over the optical - uv range , for inclinations of less than 60@xmath191 .
in particular , a uv - hard spectrum remains uv - hard .
consequently black - body spectra do appear adequate for the analysis presented here , and it is likely that the dominant uncertainty is actually in the reddening corrections , not the local emergent spectra .
the optical and uv spectral energy distributions of bhxrts in outburst have the potential to inform us of the temperature distribution of the outer disk , and hence test models for the heating of these disks .
we have collated all the spectra published to date on short period typical systems which we believe are useful in this regard . in spite of a 25year baseline ,
these data are rather limited . _ hst _ data are by far the most suitable for this work in both the uv and the optical .
ground - based optical studies suffer from uncertainties in the flux calibration , especially at short wavelengths , and from telluric absorption at long wavelengths .
consequently even from the small sample available , only a subset are really reliable .
even with high quality data , further complications arise from uncertainty in the line - of - sight extinction curve .
if the curve is known , then the reddening can be determined to high - precision using the 2175 feature . without _ a priori _ knowledge of the extinction curve , however , the relative strength of the feature is not known and the reddening can not be precisely determined .
further difficulties arise from variations in other aspects of the uv extinction curve .
the far - uv rise seems particularly variable in our sample and several of the sources considered exhibited kinks in the far - uv which could be attributed to an incorrect correction of the far - uv rise .
these further compound the uncertainty in the dereddened spectral energy distributions . in spite of these difficulties
, we can draw some useful conclusions .
provided a suitable sample of typical short - period bhxrts are selected , the optical spectra form a relatively homogeneous sample .
all have quasi - power - law shapes , with spectral slopes ( in @xmath194 vs. @xmath195 ) of 0.51.5 .
where the data are good enough to discriminate , the optical spectra tend to soften as the outburst decays , corresponding to the cooling of the outer disk .
the very homogeneity of the optical spectra means that they are of limited value alone in testing disk models , and they can not discriminate between different forms of disk heating .
it is also worth remarking that since the optical lies on the long wavelength tail of the spectral energy distribution the @xmath173 band flux is not necessarily related in a straightforward way to the irradiating luminosity and this may complicate attempts to compare optical and x - ray lightcurves ; for example the optical brightness _ at a fixed mass flow rate _ , and hence the optical to x - ray flux ratio , will depend upon the size of the hot region of the disk . to perform more useful tests requires uv data .
ideally full optical - uv coverage can be used , but the uv alone has some power to discriminate .
all spectra flatten in the uv , although depending on the source and epoch this effect may be a modest reduction in slope , or a complete turnover ( i.e. decreasing @xmath194 ) .
uncertainties in dereddening the spectra complicate the interpretation of the uv slope , but in xte j1859 + 226 evolution was seen from one form to the other , indicating that there are real variations . in this case , the evolution was in the opposite sense to the optical ; the uv spectrum hardened during the outburst decay .
these spectral shapes can be thought of as defining optical - uv spectral states in bhxrts , analogous to x - ray spectral states .
the case where the uv spectrum continues to rise can be termed the uv - hard state , and both a 062000 and gs 1124684exhibit this state .
spectra which peak and fall off in the uv can be termed uv - soft , such as those of gro j0422 + 32 . in this terminology ,
xte j1859 + 226 exhibited a transition from the uv - soft to uv - hard state as it decayed .
these spectra can be interpreted in terms of simple black body disk models .
the full range can be accommodated with two forms of heating , one varying as @xmath187 and one as @xmath186 .
the first is readily interpreted as due to irradiation by a compact x - ray source and produces uv - soft spectra . the second is usually associated with viscous heating , but can also arise if the disk is illuminated by a vertically extended x - ray source .
evolution from the first form to the second as the disk cools results in exactly the behavior seen in xte j1859 + 226 the optical spectrum softens , but the uv hardens .
none of the other sources exhibit such a clear transition , although gro j0422 + 32 exhibited similar optical spectral evolution to xte j1859 + 226suggested it might have shown the same effect if suitable late - epoch data were available . a 062000 and gs 1124684 suggest @xmath186 heating , whereas gro j0422 + 32 can be interpreted as @xmath187 irradiated heating .
it might be significant that both gro j0422 + 32 and xte j1859 + 226(at least early in the outburst ) were identified by @xcite as particularly hard sources ; it may be that there is a correlation between the optical - uv state ( i.e. the form of heating of the outer disk ) and the x - ray state .
evidence for this effect is provided by @xcite who argue that the transition to the low / hard state near the end of the outburst is accompanied by a substantial enhancement in the reprocessing efficiency .
obviously a larger sample is needed to robustly test these issues .
ideally we would like to see more examples of transitions in the optical - uv spectral state .
the interpretation of the uv - hard state remains uncertain .
these spectra are well fitted by a viscously heated shakura - sunyaev disk spectrum . in the case of a 062000
this interpretation is quite sensible assuming a lobe - filling disk accreting at 30% of the eddington rate quite naturally produces both the optical fluxes and the spectral shape without invoking reprocessing as the dominant source of optical emission . to explain gs 1124684 in the same way
, however , appears to require super - eddington accretion ( c.f.@xcite ) .
irradiation by an extended x - ray source offers a way out of this difficulty , and could bypass self - shielding of the disk @xcite , but creates its own problems by requiring a very extended x - ray emission ( or scattering ) region .
further progress in this area undoubtedly requires more observations of the quality of those of xte j1859 + 226 . just as important , however , may be for the observations to make contact meaningfully with simulations of disk outbursts .
progress is being made ( e.g.@xcite ) but simulations have not yet adequately accounted for disk warping , or realistic x - ray irradiation @xcite ; the x - ray emission geometry ( compact vs.extended ) , and spectral changes may have a significant impact on the strength of irradiation ( e.g. @xcite ) .
i am grateful for funding from nasa through hubble fellowship grant # hf-01150.01-a awarded by stsci , which is operated by the aura , inc .
, for nasa , under contract nas 5 - 26555 .
i would like to thank chris shrader , nicola masetti , and massimo della valle for providing digital optical spectra presented here ; dawn gelino for many of the distances ; erik kuulkers for helpful discussions on a 062000 ; and ivan hubeny and richard wade for discussions of disk spectral modeling .
particular thanks are due to carole haswell for originally prompting me to investigate this issue , much fruitful collaboration in this and related areas , and specifically on the data on xte j1859 + 226 included here , and to rob robinson for careful reading of the manuscript .
finally i am grateful to the referee , guillaume dubus , for constructive criticism and correction on a number of points .
some of the data included in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) or the eso / st - ecf archive facility .
stsci is operated by the association of universities for research in astronomy , inc . under nasa contract nas5 - 26555 .
specifically , this work includes observations made with the nasa / esa hubble space telescope and the nasa / esa / serc international ultraviolet explorer .
support for mast for non - hst data is provided by the nasa office of space science via grant nag-7584 and by other grants and contracts .
_ iue _ data were obtained under proposals xbmcs ( gs 1124684 , pi shrader ) , matoo , mitoo , n1205 ( gs 1124684 ) and od45z , xrocs ( gro j0422 + 32 , pi shrader ) .
_ hst _ data were obtained under proposals 3232 ( gs 1124684 , pi panagia ) and 8245 ( xte
j1859 + 226 , pi haswell ) .
this work has made extensive use of the nasa astrophysics data system abstract service .
we can explore different irradiation regimes qualitatively with a simple geometric model of reprocessing , similar to that used by @xcite for active galactic nuclei . consider a disk element at radius @xmath196 , height @xmath197 , and local slope @xmath198 .
the element is illuminated by an isotropically emitting x - ray point source at height @xmath199 above the center of the disk .
the flux @xmath200 incident upon the surface element depends upon both the distance from the x - ray source and the angle of incidence of the x - rays . in general , @xmath201 where the distance of the x - ray source from the element is @xmath202 in cylindrical coordinates , and the normal to the surface is @xmath203 . for a geometrically thin disk ( @xmath204 ) , compact x - ray source ( @xmath205 ) , and shallow disk gradients ( @xmath206 ) , this result simplifies to : @xmath207 the simplest geometry is a flat disk ( @xmath208 ) . in this case , @xmath209 ; this is the simple ` lamppost ' case referred to above .
the same would apply if the disk is not flat , but @xmath210 is constant , since then @xmath211 .
somewhat more realistic cases can be approximated with a power - law disk : @xmath212 , where @xmath213 and @xmath214 refer to values at the outer disk edge . in this case , @xmath215 @xmath216 is the case described above , and separates the behavior into two regimes .
firstly consider a concave disk , @xmath217 . if @xmath218 then the curvature of the disk is negligible and @xmath209 as for a flat disk . if @xmath219 then the vertical elevation of the x - ray source is negligible and we obtain the classical behavior of a concave disk illuminated by a central point source , with @xmath220 in the limiting case of infinitesimal curvature ( i.e. @xmath221 ) .
the dividing line between the two thus depends upon both the height of the disk relative to the height of the x - ray source and the amount of curvature of the disk . for an extended x - ray source , radius @xmath199
, we might then expect a uv - hard spectrum ( corresponding to @xmath3 ) if the x - ray source has size larger than @xmath222 .
next consider a convex disk , @xmath223 . in this case one
_ never _ expects heating to vary as @xmath187 .
if @xmath224 then the x - ray source is below the local horizon of the surface element and no irradiation is possible . if @xmath225 then again disk curvature is negligible and @xmath226 . in the intermediate cases the radial dependence is even steeper .
the caveat to this occurs if there is also local scattering material above the disk ( i.e. the disk atmosphere ) which can scatter irradiation over the disk horizon and illuminate regions that would otherwise be obscured . in this case
, @xmath187 heating remains possible . | we compile optical and uv spectra of a sample of ` typical ' short period black hole x - ray transients in outburst .
we also survey determinations of interstellar extinction and distance in order to deredden spectra and compare absolute fluxes .
hence we perform a comparative study of the broad - band spectral energy distributions ( sed ) .
we find that given such a homogeneous sample of typical sources , the optical seds form a relatively uniform set , all exhibiting quasi - power - law spectra with @xmath0 , where @xmath1 ( steeper than the canonical @xmath2 disk spectrum ) .
all become flatter in the uv , although there is more diversity here .
the seds studied can be broadly divided into two optical - uv spectral states .
the uv - hard spectra , e.g. a 062000 and gs 1124684 , continue to rise in the far - uv .
the uv - soft spectra , e.g.gro j0422 + 32 , drop off .
xte j1859 + 226 evolved from uv - soft to uv - hard as it decayed indicating that this effect is a real difference , not a dereddening artifact .
all of the spectra can be fitted by a generalized black body disk model with two forms of heating , resulting in the two states .
the uv - soft state is consistent with a disk illuminated by a central point source , with irradiative heating dominating over viscous .
the uv - hard state is well described by a viscously heated disk , although this requires very high mass flow rates in the case of gs 1124684 .
alternatively , a uv - hard spectrum can be produced if the disk is illuminated by a vertically extended x - ray source such as a central scattering corona or jet .
since scattering is assumed by some numerical simulations , it is worth emphasizing that where illumination comes from ( non - local ) scattering high above the disk , we generically expect a steeper radial dependence of x - ray heating ( @xmath3 ) than is usually assumed ; it is this steep dependence that leads to the uv - hard spectrum . |
0.2 cm _ ultra - relativistic solution_0.3 cm in analytical approach @xcite , the ultra - relativistic approximation is adopted for @xmath37 , the term @xmath38 in eqs .
( [ ppeq],[en1],[ppeqp],[en1p ] ) is neglected .
it turns out that approximated thomas - fermi equations are the same for both cases of neutral and charged cores , and solution @xmath39 , @xmath40^{-1 } , & { \rm for}\quad x<0 ,
\\ \frac { \sqrt{2}}{(x+1.89 ) } , & { \rm for}\quad x>0 , \end{array}\right\ } , \label{popovs2}\ ] ] where @xmath41 and the pion compton length @xmath42 . at the core center @xmath43 , @xmath44 . on core surface @xmath45 , @xmath46 , indicating that the ultra - relativistic approximation is applicable for @xmath47
this approximation breaks down at @xmath48 . clearly , it is impossible to determine the value @xmath30 out of ultra - relativistically approximated equation , and full thomas - fermi equation ( [ eposs ] ) with source terms eq .
( [ en1 ] ) for the neutral case , and eq .
( [ en1p ] ) for the charged case have to be solved . for @xmath49 where @xmath50 , we treat the term @xmath38 in eqs .
( [ en1],[en1p ] ) as a small correction term , and find the following inequality is always true @xmath51 where @xmath52 and @xmath53 stand for electron densities of neutral and super charged cores . for the range @xmath34 , @xmath54 in the case of super charged core , while @xmath55 in the case of neutral core , which should be calculated in non - relativistic approximation : the term @xmath56 in eq .
( [ en1 ] ) is neglected . in conclusion ,
the physical scenarios and thomas - fermi equations of neutral and super charged cores are slightly different .
when the proton density @xmath7 of cores is about nuclear density , ultra - relativistic approximation applies for the coulomb potential energy @xmath57 in @xmath58 and @xmath59 , and approximate equations and solutions for electron distributions inside and around cores are the same . as relativistic regime @xmath60 and non - relativistic regime @xmath61 ( only applied to neutral case ) are approached , solutions in two cases are somewhat different , and need direct integrations . and
electron mass - gap @xmath62 in the unit of @xmath63 are plotted as a function of @xmath64 .
the potential depth inside core ( @xmath65 ) is about pion mass @xmath66 and potential energy - gap and electron mass - gap are indicated .
the radius @xmath30 where electron distribution @xmath67 vanishes in super charged core case is indicated as @xmath68 , since it is out of plotting range.,width=377,height=226 ] | based on the thomas - fermi approach , we describe and distinguish the electron distributions around extended nuclear cores : ( i ) in the case that cores are neutral for electrons bound by protons inside cores and proton and electron numbers are the same ; ( ii ) in the case that super charged cores are bare , electrons ( positrons ) produced by vacuum polarization are bound by ( fly into ) cores ( infinity ) .
0.2 cm _ equilibrium of electron distribution in neutral cores._0.3 cm in refs .
@xcite , the thomas - fermi approach was used to study the electrostatic equilibrium of electron distributions @xmath0 around extended nuclear cores , where total proton and electron numbers are the same @xmath1 .
proton s
density @xmath2 is constant inside core @xmath3 and vanishes outside the core @xmath4 , @xmath5 where @xmath6 is the core radius and @xmath7 proton density .
degenerate electron density , @xmath8 where electron fermi momentum @xmath9 , fermi - energy @xmath10 and coulomb potential energy @xmath11 are related by , @xmath12^{1/2}-m_ec^2 - v_{\rm coul}(r ) .
\label{eeq0}\end{aligned}\ ] ] the electrostatic equilibrium of electron distributions is determined by @xmath13 which means the balance of electron s kinetic and potential energies in eq .
( [ eeq0 ] ) and degenerate electrons occupy energy - levels up to @xmath14 . eqs .
( [ en],[eeq0],[eeq ] ) give the relationships : @xmath15^{1/2 } ; \label{ppeq}\\ n_e(r ) & = & \frac { 1}{3\pi^2(c\hbar)^3}\left[v^2_{\rm coul}(r)+2m_ec^2v_{\rm coul}(r)\right]^{3/2}. \label{en1}\end{aligned}\ ] ] the gauss law leads the following poisson equation and boundary conditions , @xmath16;\quad v_{\rm coul}(\infty)=0,\quad v_{\rm coul}(0)={\rm finite}. \label{eposs}\end{aligned}\ ] ] these equations describe a thomas - fermi model for neutral nuclear cores , and have numerically solved together with the empirical formula @xcite and @xmath17-equilibrium equation @xcite for the proton number @xmath18 and mass number @xmath19 , where @xmath20 is the neutron number . |
many studies of tmx s ( tm = transition metal and x = c or n ) are motivated by a curiosity on the properties of tmx s , such as their mixture of covalency , ionicity , and metallicity , and by a suggested importance for heterogeneous catalysis . the attention to tmx s
as potential catalysts started with an observation by levy and boudart that the tmc s show a pt - like behavior in several catalytic reactions.@xcite according to more recent investigations , `` early tmc s and tmn s often demonstrate catalytic advantages over their parent metals in activity , selectivity and resistance to poisoning '' and `` for several reactions , such as hydrogenation reactions , catalytic activities of tmc s and tmn s are approaching or surpassing those of group viii noble metals''.@xcite the tmx surfaces are also technologically important as substrate materials in growth processes , _
e.g. _ , in wear - resistant multilayer coatings on industrial cutting tools @xcite and for growth of carbidic nanostructures.@xcite for catalytic applications , the stable or ideal surfaces are not necessarily the most suitable ones . often the best site for a reaction
is found on a stepped or in some way non - perfect surface , _
e.g. _ , at kinks or around defects , where less stable faces of the material are exposed .
such sites often host surface states or surface resonances ( sr s ) .
this calls for studies on surfaces that present such surface states or resonances . in previous studies , the reactivities of the tic(@xmath0 ) and tin(@xmath0 )
surfaces are attributed to the presence of sr s of both ti and c / n character.@xcite calculated trends in atomic adsorption strength are explained with a concerted - coupling model ( ccm ) , in which the atomic frontier orbitals interact with both types of sr s .
more recently , we have indentified a descriptor , defined as the mean energy of the tm - derived sr ( tmsr ) , for atomic and molecular adsorption and for activation - energy barriers.@xcite hence , the existence of several linear relations between the atomic and molecular adsorption energies , `` scaling relation '' , have been shown.@xcite such relations are of importance in the design of novel types of catalysts .
@xcite this paper is devoted to a deeper and more generalized understanding of the chemisorption on the tmc(@xmath0 ) surfaces , thus also laying the ground for the above mentioned descriptor .
we extend the work done in our previous study of the atomic adsorption on tic(@xmath0 ) and tin(@xmath0 ) and use density - functional theory ( dft ) to investigate whether the ccm is applicable to other tmc s .
our method consists of a detailed study of the trends in reactivity along periods and groups of the substrate parent metal and of the adsorbate , correlated with a careful mapping and analysis of the underlying details of the changes in surface electronic structure upon adsorption .
the substrates chosen in our study are the tmc s formed with the parent metals sc , ti , v , zr , nb , mo , ta , and w in nacl structure ( see fig .
[ fig : def_tmc ] ) . as adsorbates
we choose atomic h , b , c ,
n , o , and f as well as the molecules nh , nh@xmath2 and nh@xmath3 . these particular choices of tmc s and
adsorbates allow us to capture changes in adsorption properties along both periods and groups in the periodic table .
all studied tmc s adopt a nacl structure either in stable or in metastable phase .
the early transition - metal carbides under investigation.,scaledwidth=25.0% ] experiments show that after heating and ion or electron bombardment under low - temperature and low - pressure conditions the ( @xmath0 ) surfaces of tic,@xcite vc,@xcite zrc,@xcite nbc,@xcite and tac@xcite are unreconstructed and tm terminated . therefore , our study deals with unreconstructed and tm - terminated ( @xmath0 ) tmc surfaces .
surface electronic structure characterizations with angle - resolved photoemission studies ( arpes ) show the presence of surface states on the ( @xmath0 ) surfaces of tic , @xcite zrc , @xcite nbc,@xcite and tac.@xcite also , experiments show that both h@xmath2 and o@xmath2 dissociate on tic(@xmath0),@xcite zrc(@xmath0),@xcite and nbc(@xmath0).@xcite despite this large number of experimental studies , theoretical investigations on tmc(@xmath0 ) surfaces are scarce.@xcite this paper presents our results in the following order .
first , a trend study of the clean tm - terminated tmc(@xmath0 ) surfaces is presented in section [ sec : surface ] . in particular
we address the properties of the surface electronic structures that are relevant for the adsorption mechanisms .
then , in section [ sec : ads ] the adsorption - energy trends are presented together with the trends in the adorption - induced changes in surface electronic structure .
the different trends in adsorption - energy and electronic - structure changes are analyzed and discussed in section [ sec : discussion ] , thus connecting our results to the previously proposed ccm .
the paper is concluded by section [ sec : conclusion ] , which summarizes the main conclusions of our investigation and makes some outlooks to possible ramifications and further investigations .
in this section the computational details and results from our dft calculations on the clean tm - terminated tmc(@xmath0 ) surfaces are presented .
first , the stability of the ( @xmath0 ) surfaces is considered by comparing the cleavage energies of these surfaces with the corresponding results for the ( @xmath4 ) surfaces .
then , the relaxed surface structures are presented and compared with existing results .
this is followed by a charge transfer analysis and a detailed analysis of both the energy- and space - resolved surface densities of states ( dos s ) .
particular emphasis is put on the presence and character of sr s on the tmc(@xmath0 ) surfaces . throughout the presentation ,
analyses of the trends with respect to the tm component of the studied tmc s are made .
the surface calculations presented in this paper are performed within the dft formalism using the plane - wave pseudopotential code dacapo.@xcite the ion - electron interaction is treated with vanderbilt ultrasoft pseudopotentials.@xcite the exchange - correlation energy is included by the generalized gradient approximation ( gga ) using the pw91 functional.@xcite we utilize a slab geometry , with slabs of @xmath5 to @xmath6 bilayers ( a bilayer being a unit of one tm layer and one c layer ) , a vacuum region thickness corresponding to @xmath7 bilayers , that is , at least @xmath8 , and periodic boundary conditions . each atomic layer is composed of one atom in a ( @xmath9 ) geometry .
the atoms in the three ( four ) outermost atomic layers on the tm - terminated side of a @xmath5 ( @xmath10 ) bilayer thick slab are allowed to relax until the sum of the remaining forces on all relaxed atoms is less than @xmath11 ev / , while the remaining atomic layers are fixed at the bulk geometry .
a monkhorst - pack sampling @xcite of @xmath12 special @xmath13-points and a plane - wave energy cutoff of @xmath14 ev are used . the slab used to model
the ( @xmath0 ) surface is asymmetric , which gives rise to a discontinuity in the electrostatic potential at the cell boundary .
this is corrected for by using the scheme in ref . .
to characterize the surfaces we utilize several electronic structure tools .
a bader analysis is used to calculate the charge localization around individual atoms.@xcite total and local , that is , atom - projected , dos s for the surface bilayer are obtained by projecting the kohn - sham wave functions onto individual atomic orbitals and plotted as a function of energy ( relative to the fermi level @xmath15 ) . to identify the surface specific properties that arise upon creation of the surface , the surface dos s
are compared to the bulk dos s by studying the differences between the two quantities .
information about the spatial localization of the surface - localized states is provided by analyzing the space - resolved surface dos , that is , the kohn - sham wave functions .
the cleavage energy @xmath16 , that is , the energy needed to create two surfaces upon cleavage of a bulk structure along a specific crystallographic plane , is calculated as @xmath17 where @xmath18 is the total energy of a slab with @xmath19 tmc bilayers that exposes the two surfaces under investigation and @xmath20 is the bulk energy of one tmc bilayer , if @xmath19 is sufficiently large .
thus , the cleavage energy corresponds to the sum of the surface energies of the two surfaces obtained upon cleavage . in the case of calculations on stoichiometric tmc(@xmath0 ) slabs ,
the cleavage energy is equal to the sum of the surface energies of the tm - terminated surface and of the c - terminated surface . our calculated @xmath16 values for the tmc(@xmath0 ) surfaces , after relaxation of only the tm - terminated side of the slabs ,
are given in table [ tab : tmc_surf_relax ] .
along a period @xmath16 shows a maximum for group iv , thus showing the same variations as the ones found for the bulk cohesive energies in our previous study.@xcite down a group , the variations in @xmath16 are small but discernible and do not show any apparent correspondence to the bulk cohesive energies of ref . .
according to the calculated @xmath16 values , tic(@xmath0 ) is the surface that requires the most energy to create among the considered tmc surfaces .
nevertheless , this surface is routinely grown by chemical - vapor deposition ( cvd ) under high temperatures as wear - resistant coating on industrial cutting tools.@xcite the calculated @xmath16 value for the tic(@xmath0 ) surface agrees well with those of previous dft calculations.@xcite compared to the tmc(@xmath4 ) surfaces , the tmc(@xmath0 ) surfaces are found to have a higher @xmath16 value ( see table [ tab : tmc_surf_relax ] ) and are thus less stable . also , the variations in @xmath16 are larger for tmc(@xmath0 ) than for tmc(@xmath4 ) surfaces .
both these properties can be attributed to the polar nature of the ( @xmath0 ) surface.@xcite [ cols="^,^,^,^,^,^,^,^,^,^,^",options="header " , ] in this section we present results for molecular adsorption on the vc(@xmath0 ) surface , which is here chosen as a prototype for molecular adsorption on the tmc(@xmath0 ) surface .
the molecular adsorbates are the nh@xmath21 molecules ( @xmath22 ) , which are a part of our study in ref . .
the same supercell size and computational parameters as those used for the atomic adsorption are employed for these calculations ( see section [ sec : surf_comp ] ) .
the molecules are adsorbed with the n atom closest to the surface in the fcc site ( the stable site for n atomic adsorption ) .
the adsorption energy is calculated relative to the energy of an isolated nh@xmath21 molecule as @xmath23 the charge transfer between substrate and absorbed molecule is calculated as the sum of the bader charges on each of the constituent atoms in the molecule . a summary of the results is given in table [ tab : nhx_on_vc ] .
as the molecule size @xmath24 increases , the perpendicular distance to the surface increases , while both the adsorption energy @xmath25 and the charge transfer from surface to molecule decrease .
figure [ fig : vc_deltados_nhx ] shows calculated ldos s ( projected on the molecule atoms ) and @xmath26dos
s ( projected on the substrate atoms ) for n and nh@xmath21 ( @xmath22 ) on vc(@xmath0 ) . the energy and structure of the ldos s vary between the different adsorbates . for atomic n
, it consists of one delocalized region in the middle of the uvb . for nh ,
there is a slightly weaker delocalized region of n character in the middle of the uvb , together with a sharp peak of mixed n and h character in the lower part of the uvb . for nh@xmath2 ,
the delocalized region of n character in the middle of the uvb is further weakened and a new sharp peak of mixed n and h character appears below the uvb . for nh@xmath3 , the delocalized region in the middle of the uvb disappears , while the sharp peak below the uvb is stronger and lies at lower energy . for all adsorbates
, there is a quenching of both the tmsr and the csr s .
these sr depletions are largest for n and decrease every time a h atom is added to the molecule . hence
, each addition of a h atom weakens the interaction between the adsorbed molecule and the sr s .
as described above , we find several trends in the calculated adsorption energies : `` m''-shaped variations with respect to atomic adsorbate , linear variations with respect to substrate , qualitatively different types of adsorption on scc and for f adatom , and trends in molecular adsorption strength . in this section ,
we describe how these trends can be described within our previously proposed concerted - coupling model ( ccm ) for atomic adsorption on tix(@xmath0 ) surfaces.@xcite first , the ccm is described by giving an account of the main results of refs . and shown to apply also on the here considered tmc s by pointing out the similarities in electronic structure results .
then , the different adsorption - energy trends are explained in terms of the ccm .
finally , the results lay the ground for the single descriptor @xmath27 for the adsorption strength , recently introduced in ref . . for tm surfaces ,
the @xmath28-band model yields a successful description of electronic structure and adsorption.@xcite for instance , its key parameter @xmath29 , the mean energy of the substrate @xmath28 band , is a good descriptor for , _
e.g. _ , adsorption.@xcite such a fact facilitates the design of new tm catalysts by computational screening.@xcite for adsorption on tmc s , however , there are deviations from the _
d_-band model.@xcite indeed , we find that different tmc surfaces can have the same value of @xmath29 but different @xmath25 values .
for example , the @xmath29 values for tic(@xmath4 ) and tic(@xmath0 ) surfaces are as close as @xmath30 ev and @xmath31 ev , whereas the @xmath25 values for oxygen differ considerably , being @xmath32 ev and @xmath33 ev , respectively.@xcite in the _
d_-band model , adsorption trends are explained by the interaction between the adatom frontier orbital and the narrow tm @xmath28 band . using the terminology of the newns - anderson ( na ) model for chemisorption ,
@xcite this bond is typically `` strong '' and results in the formation of separated bonding and antibonding adatom substrate states ( see case d in fig . [
fig : na]).@xcite in refs . , and , atomic adsorption on the tic and tin ( @xmath0 ) surfaces is described as a result of two types of interactions between adsorbate and substrate . in the terminology of the na model , the coupling of the adatom frontier orbital is typically `` strong '' with the substrate tisr and `` weak '' with the substrate xsr s ( x = c or n ) ( as in case d and in cases a c in fig .
[ fig : na ] , respectively ) .
a concerted action of these couplings gives a qualitative explanation for the calculated adsorption energy trends on the tic(@xmath0)@xcite and tin(@xmath0)@xcite surfaces .
the tisr is present on the tix(@xmath0 ) surfaces but not on the ( @xmath34 ) ones .
the xsr s are found in the tix(@xmath0 ) uvb s . in
the first mentioned coupling above , the large overlap of the localized tisr with the adatom orbital causes a strong adatom tisr interaction ( in the na sense ) .
well - separated bonding and antibonding states of mixed adsorbate and ti character are then formed ( case d in fig .
[ fig : na ] ) .
the bonding - state energy lies below the free - adatom and tisr levels , while the antibonding state resides above the tisr level . in the other coupling , the bonding adatom
tisr level interacts with the xsr s present in the substrate uvb . due to the short range of the xsr s ( compared to the tisr ) , this interaction is weak ( in the na sense ) and causes a mixture of broadening and shifting of the bonding adlevel tisr state . a state located in the middle of the uvb is mainly broadened ( case b in fig .
[ fig : na ] ) , while a state at the edge of the uvb is mainly shifted away from the uvb center of mass ( cases a and c in fig .
[ fig : na ] ) .
evidence for this concerted coupling is given by detailed analyses of the calculated adsorbate - induced dos s ( @xmath26dos s ) and of real - space visualizations of the kohn - sham wave functions .
these show that upon adsorption , there are ( i ) a sharp decrease in dos at the tisr energy ; ( ii ) a sharp increase of dos just above the tisr energy ; ( iii ) depending on the adatom species , narrow or broad bands of mainly adatom character at the edge of or within the substrate uvb energy range , respectively ; ( iv ) a depletion of x - localized uvb states at the xsr energies ; and ( v ) a formation of adatom - localized sub - peaks in between the energies of the substrate xsr s . points ( i)(iii ) show that bonding and antibonding adlevel tisr states are formed , while points ( iii)(v ) prove a coupling between the bonding adlevel tisr state and the xsr s in the substrate uvb .
more detailed analyses of the trends in calculated @xmath26dos s and kohn - sham wave functions for second- and third - period adatoms on tix(111 ) surfaces show that ( i ) the magnitude of the dos reduction at the tisr energy decreases successively as the adatom number @xmath35 increases along a period and ( ii ) the adatom
x bonding character of the adatom - localized peaks increases as the adatom number @xmath35 increases along a period .
these trends , which arise from the successive lowering of the adlevel energy as @xmath35 increases , indicate that the adatom
tisr coupling decreases in strength along each adatom period , towards the right in the periodic table , while the contribution of the xsr s to the bonding increases .
such trends provide a basis for understanding the calculated trends in atomic adsorption energies .
for instance , the maximum for the group - vi adatoms ( o and s ) is explained in terms of a stronger coupling to the substrate csr s . on the other hand ,
the weaker bonding of group - vii adatoms ( f and cl ) is explained by a weakened coupling to the csr s , due to the adatom state being almost fully ionized by its interaction with the tisr .
thus , the f and cl adsorbates lack almost any covalent interaction with the substrate uvb and their adsorption is practically ionic in nature , as confirmed by bader and charge - density analyses .
also , the almost equal adsorption strengths of c and n can be explained in terms of the opposite trends in adatom tisr and adatom csr coupling strengths found for varying @xmath35 values . on the same basis ,
the ccm is also able to describe changes in adsorption energies arising from changes in adsorption site , _ i.e. _ , between fcc , hcp , and top o adatoms on tic(@xmath0).@xcite the calculated energetics and electronic structures of atomic adsorption on the tmc(@xmath0 ) surfaces described in section [ sec : ads ] can be analyzed in the same way as described above for tix(@xmath0 ) .
the fact that upon adsorption the tmsr s and csr s are quenched on all considered surfaces shows that both types of resonance participate in the substrate
adatom bond .
the strong character of the adatom
tmsr bonds is evidenced by the formation of well - separated bonding and antibonding states of mixed adatom and tm character , at energies below and above the tmsr s , respectively .
the weak character of the adatom
csr bonds is confirmed by the broadening and/or shift of the bonding adatom
tmsr level .
the exception is scc(@xmath0 ) , on which the adatom
csr coupling approaches a strong na character , with a strong depletion of csr states in the middle of the uvb and the formation of separated bonding and antibonding states below and above the uvb , respectively .
this is due to the stronger csr s on this surface , which in turn arises from the stronger c c bonds in bulk scc than in the other tmcs.@xcite also , the adatom tmsr coupling is qualitatively different on scc(111 ) , compared to the other tmc s , as the tmsr of clean scc(111 ) lies above @xmath15 and is thus empty before adsorption .
thus , the basic adsorption mechanism is the same on all the considered tmc(@xmath0 ) surfaces , implying that the general applicability of the ccm includes tmc(@xmath0 ) surfaces .
in addition , scc(111 ) provides a playground for testing the contribution of the adlevel csr coupling to the total adsorption strength . in the following ,
we give more explicit support to the applicability of the ccm to the tmc(@xmath0 ) surfaces by showing that the ccm can be used to understand the major features of the calculated atomic and molecular adsorption - energy trends on the considered tmc(@xmath0 ) surfaces .
figure [ fig : ads_on_tmx_all_ads_sub](b ) shows the calculated energy values for atomic adsorption on tmc(@xmath0 ) surfaces as functions of the substrate tm group number . in this subsection
we show that the ccm can capture its main trends . in a simple molecular two - level picture ,
the strength of a bond is related to the energy shift between the levels of the isolated constituents and of the hybridized states resulting from the coupling . in a simplified version of the newns model , an adsorbate and the active part of the substrate
can be viewed as such a `` molecule''.@xcite accordingly , from the calculated @xmath26dos s the bond strength can be extracted as the energy difference between the final bonding adatom level ( _ i.e. _ , after coupling to both tmsr and csr s ) and the tmsr peak in the surface dos .
the ccm is then able to explain the trends in bond strength by relating this energy difference to concerted adatom tmsr and adatom csr interactions . in the following ,
we employ such an approach to understand the qualitatively different trends in @xmath25 values for c and o adatoms , as the substrate tm group number increases from iv to vi [ fig .
[ fig : ads_on_tmx_all_ads_sub](b ) ] . for c , the @xmath25 values are approximately constant between groups iv and vi . the calculated @xmath26dos s ( see fig . [
fig : c_deltados_all_substrates ] ) show that the final bonding adatom level lies at the upper edge of the uvb on all considered substrates .
thus on all these substrates , the energy difference between the final adatom level and the original substrate tmsr is approximately constant .
this implies that the result of the concerted action of the adatom tmsr and adatom
csr s interactions is approximately constant .
in particular , fig .
[ fig : c_deltados_all_substrates ] shows that the dos s of all substrates resemble the same na type of adlevel
csr s interaction , that is , type c in fig .
[ fig : na ] .
for o , with its lower @xmath36 level , the @xmath25 values decrease strongly as the substrate tm group number increases from iv to vi .
this is reflected in the calculated @xmath26dos s ( fig .
[ fig : o_deltados_all_substrates ] ) by a strongly decreasing energy difference between the final adatom level and the original substrate tmsr peak .
thus , in contrast to c , for o the concerted action of the two types of coupling yields qualitatively different results . in particular , fig .
[ fig : c_deltados_all_substrates ] shows dos s that correspond to adlevel
csr s interactions that vary in a qualitative way from ( _ _ cf .
_ _ fig .
[ fig : na ] ) na case a ( on group iv ) , to na case b ( on group v ) , and to na case c ( on group vi ) . as a consequence , the adlevel
csr s interactions cause a downward energy shift ( on group iv ) , a broadening ( on group v ) , and an upward energy shift ( on group vi ) of the bonding adlevel .
we also notice that for both adsorbates the antibonding adatom
tmsr level lies above @xmath15 ( and is thus empty ) for all substrates except the group - vi tmc s @xmath1-moc and wc .
this filling of antibonding states implies an extra decrease in adatom
tmc bond strength , as group v @xmath37 vi .
indeed , this can be seen in the calculated @xmath25 trend for c [ fig .
[ fig : ads_on_tmx_all_ads_sub](b ) ] . for all adsorbates except f
, the calculated @xmath25 values increase from scc(@xmath0 ) to tic(@xmath0 ) . as described above , the adlevel
csr s interaction is more pronounced on scc(@xmath0 ) than on the other tmc s , thus approaching the strong na limit . also , the adatom tmsr coupling is qualitatively different on scc(@xmath0 ) , due to the tmsr of clean scc(@xmath0 ) lying above @xmath15 and being thus empty before adsorption .
the weaker adsorption on scc(@xmath0 ) can thus be understood to be due to the qualitatively different interaction with the two types of sr s .
the adsorption - energy trends with respect to adsorbate [ fig .
[ fig : ads_on_tmx_all_ads_sub](a ) ] show
m-shaped @xmath25 trends for most of the considered tmc s .
the trend on tic(@xmath0 ) was analyzed in ref . and explained within the ccm to arise from competing opposite trends in adsorbate
tmsr and adsorbate
csr s interactions . the calculated @xmath26dos s for the here considered tmc s ( illustrated for vc in fig .
[ fig : vc_deltados_all_adsorbates ] ) show trends that are similar to those found for tic . at the same time , the @xmath25 trends for all considered tmc s resemble the one on tic(@xmath0 ) , with a local maximum for the o adatom
. however , on group - v tmc s the calculated @xmath25 values for o are approximately equal to the ones for c , and on group - vi tmc s the o values are smaller than the ones for c. thus , the chemisorption strength of o appears to weaken relative to c as the substrate tm group number increases .
this can be understood to be a consequence of the different @xmath25 trends for c and o that were described in sec .
[ sec : disc_substrate ] above , with the adsorption strength decreasing much faster for o than for c as the tm group number increases from iv to vi .
on the other hand , the @xmath25 values for scc(@xmath0 ) show a monotonic increase between b and o adatoms . as described above , the adlevel
csr s interaction is stronger on scc(@xmath0 ) than on the other considered tmc(@xmath0 ) surfaces .
therefore , the monotonically increasing trend ( as b @xmath37 o ) of the adlevel
csr s coupling contribution to the adatom
tmc bond is stronger on scc(@xmath0 ) .
finally , on all substrates there is a sharp decrease in adsorption energy , as o @xmath37 f. this arises from the weakened adlevel
csr s interaction , which is due to the fact that the adatom is almost fully ionic from its interaction with the tmsr , as described previously on tic(@xmath0).@xcite the adsorption energies of the considered molecular adsorbates nh@xmath21 ( @xmath38 ) on vc(@xmath0 ) show a decreasing trend as @xmath24 increases ( table [ tab : nhx_on_vc ] ) .
this trend can be understood in terms of the ccm as arising from the concerted coupling between each molecular orbital and the two types of sr s .
the quenchings of both tmsr and csr s in the calculated @xmath26dos s for nh@xmath21 on vc(@xmath0 ) ( fig .
[ fig : vc_deltados_nhx ] ) indicate the presence of couplings to both types of sr s .
in addition , the adsorbate - projected ldos s show strongly bound sharp peaks of mixed n and h character as well as more delocalized regions of only n character within the uvb .
a comparison with calculated dos s for the free nh@xmath21 molecules shows that the sharp peaks correspond to low - energy molecular levels that interact weakly with the substrate sr s . on the other hand ,
the delocalized regions arise from the coupling of higher - energy molecular levels with the sr s , in a way that is similar to the adsorption of atomic n. figure [ fig : vc_deltados_nhx ] shows that as the number @xmath24 of h atoms in the molecule increases , the interaction of the higher - energy molecular levels with the sr s weakens , thus lowering the adsorption strength .
the above discussion shows that for the atomic and molecular adsorption on the tmc(@xmath0 ) surfaces a picture based on the concerted action of two types of adatom substrate interactions applies .
it can be used to describe key features of the calculated adsorption - strength and electronic - structure trends . in this concerted - coupling model ( ccm ) , two types of sr s participate in the bond , tm - localized sr s ( tmsr ) and c - localized sr s ( csr ) .
therefore we conclude that the ccm is valid for atomic adsorption on all the considered tmc(111 ) surfaces . to be able to describe the variations in adsorption strength in a simple yet efficient way , a descriptor @xmath27 , defined as the mean energy of the tmsr , can be introduced.@xcite due to the approximately constant energy difference between tmsr and csr s in the considered compounds , such a descriptor is able to capture the important variations of both tmsr and csr energy trends while being at the same time both conceptually simple and measurable or calculable .
this is confirmed by fig .
[ fig : e_ads_vs_epsilonccm ] , which shows a linear correlation between our calculated @xmath25 values and @xmath27 for each of the considered atomic adsorbates , also when the results of ref .
are significantly extended .
the exception is scc(@xmath0 ) , which is a consequence of the mentioned qualitative difference in electronic structure of this surface , that is , an empty tmsr and a strong adlevel csr coupling .
as discussed above , the variations in the gradients of the lines for the different adsorbates can be understood within the ccm from the details in the interactions between the adlevel and the different sr s .
correlation between @xmath25 and @xmath27 for the atomic adsorbates h , b , c , n , o and f.,scaledwidth=50.0% ] in ref . , linear correlations are also found between @xmath27 and the molecular adsorption energies of nh , nh@xmath2 , and nh@xmath3 , as well as the activation - energy barriers for n@xmath2 . this is a consequence of the fact that both atomic and molecular adsorptions appear to follow the same basic mechanisms of the ccm , as argued above . as a consequence ,
we show in ref . that scaling and brnsted - evans - polanyi ( bep ) relations apply for adsorption on tmc(111 ) surfaces .
such findings are of importance for the design of new catalysts .
the possibility to understand materials has today increased considerably , as dft has developed into a predictive theory
. one obvious application of dft is to calculate numbers for bonding strengths , structure parameters and coordinates , _ etc . _
another , slightly more demanding , application is to look for more detailed information and develop conceptual frameworks and models in which we can formulate our understanding and on which we can base our further thinking , including ideas about new materials .
we have found that various densities of states ( dos s ) are excellent tools for such understanding and vehicles for thought . in our approach ,
extensive use of total , local , projected , and difference dos s are cornerstones in the building - up of a conceptual framework and model .
this study deals with understanding the atomic and molecular adsorption on the tmc(@xmath0 ) surfaces using such a detailed electronic structure analysis approach . by extensive dft calculations on scc ,
tic , vc , zrc , nbc , @xmath1-moc , tac , and wc , we study trends in clean surface properties , adsorption energies , and various dos s . in brief , we find that only a certain part of the surface - localized electronic structure , corresponding to the tm - localized ( tmsr ) and the c - localized ( csr ) surface resonances ( sr s ) are of importance to understand the trends in adsorption energies from one tmc surface to another and from one adsorbate to another . despite the tm termination of the investigated tmc surfaces , the second - layer c atoms are found to play a crucial role in the chemisorption .
this is particularly evident on the scc(@xmath0 ) surface , where the tmsr is empty and where the csr s are particularly strong .
having thus identified the key parameters , we possess the foundations for a concerted - coupling model ( ccm ) in which trends in adsorption strength are the result of a concerted action of both adsorbate tmsr and adsorbate
csr s couplings .
this has earlier been shown for the tic and tin ( @xmath0 ) surfaces@xcite but the breadth and versatility are here shown by applications to other tmc s and trends , including : adsorption trends with respect to substrate , adsorption trends with respect to atomic adsorbate , adsorption trends on scc(@xmath0 ) , adsorption trend for group - vii adatoms , and molecular - adsorption trends .
this deepens the interpretation of the model and broadens its usefulness .
the applicability of the ccm to both atomic and molecular adsorption opens up the possibility to study reactions for , _
e.g. _ , catalytic applications on this class of materials .
for example , it allows the formulation of a single descriptor @xmath39 for the adsorption strength.@xcite also , it implies the existence of scaling relations between molecular and atomic adsorption strengths as well as brnsted - evans - polanyi relations.@xcite since the ccm framework is based on rudimentary bonding principles with general applicability , we believe that it is possible to generalize it to other materials that possess surface - localized states . we have already shown its applicability to atomic adsorption on tin(@xmath0)@xcite and believe that the same chemisorption mechanism should be valid for other nitrides .
ligand and vacancy systems have been shown to belong to the group of materials where the ccm applies,@xcite as do certain tm surfaces,@xcite where , however , it does not need to replace the sufficient and natural _ d_-band model .
natural extensions should include tm oxides , sulfides , and borides .
design of materials , including atomic - scale engineering , is also an enticing prospect for further applications .
valuable discussions with anders hellman are acknowledged .
the calculations were performed at hpc2n and nsc via the swedish national infrastructure for computing .
b. i. lundqvist gratefully acknowledges support from the lundbeck foundation ( denmark ) via the center for atomic - scale materials design .
k. burke , j. p. perdew , and y. wang , in _ electronic density functional theory , recent progress and new directions _ , edited by j. f. dobson , g. vignale and m. p. das ( plenum press , new york and london 1998 ) ; pp .
81111 .
f. jona and p. m. marcus , in _ the structure of surfaces ii _ , edited by j. f. van der veen and m. a. van hove ( springer , berlin , 1988 ) p. 90 ; k. heinz , surf .
sci . * 299/300 * , 433 ( 1994 ) ; and references therein .
e. m. fernndez , p. g. moses , a. toftelund , h. a. hansen , j. i. martinez , f. abild - pedersen , j. kleis , b. hinnemann , j. rossmeisl , t. bligaard , and j. k. nrskov , angew .
int . ed . *
47 * , 4683 ( 2008 ) . | this study explores atomic and molecular adsorption on a number of early transition - metal carbides ( tmc s ) by means of density - functional theory calculations .
the investigated substrates are the tm - terminated tmc(@xmath0 ) surfaces , of interest because of the presence of different types of surface resonances ( sr s ) on them and because of their technological importance in growth processes .
also , tm compounds have shown potential in catalysis applications .
trend studies are conducted with respect to both period and group in the periodic table , choosing the substrates scc , tic , vc , zrc , nbc , @xmath1-moc , tac , and wc ( in nacl structure ) and the adsorbates h , b , c , n , o , f , nh , nh@xmath2 , and nh@xmath3 . trends in adsorption strength are explained in terms of surface electronic factors , by correlating the calculated adsorption energy values with the calculated surface electronic structures .
the results are rationalized with use of a concerted - coupling model ( ccm ) , which has previously been applied succesfully to the description of adsorption on tic(@xmath0 ) and tin(@xmath0 ) surfaces [ solid state commun . * 141 * , 48 ( 2007 ) ] .
first , the clean tmc(@xmath0 ) surfaces are characterized by calculating surface energies , surface relaxations , bader charges , and surface - localized densities of states ( dos s ) .
detailed comparisons between surface and bulk dos s reveal the existence of transition - metal localized sr s ( tmsr s ) in the pseudogap and of several c - localized sr s ( csr s ) in the upper valence band on all considered tmc(@xmath0 ) surfaces . the spatial extent and the dangling bond nature of these sr s are supported by real - space analyses of the calculated kohn - sham wave functions .
then , atomic and molecular adsorption energies , geometries , and charge transfers are presented .
an analysis of the adsorbate - induced changes in surface dos s reveals a presence of both adsorbate tmsr and adsorbate
csr s interactions , of varying strengths depending on the surface and the adsorbate .
these variations are correlated to the variations in adsorption energies .
the results are used to generalize the content and applications of the previously proposed ccm to this larger class of substrates and adsorbates .
implications for other classes of materials , for catalysis , and for other surface processes are discussed . |
quantum correlations in composite systems transcend entanglement @xcite .
a bipartite quantum state @xmath0 can be defined as nonclassical or nonclassically correlated if it can not be expressed as a convex mixture of local basis states of subsystems @xmath1 and @xmath2 @xcite .
consequently , all inseparable ( entangled ) states as well as the majority of separable states are nonclassical .
general nonclassical correlations , however , can be mapped to entanglement in a very precise sense , which provides an insightful framework for their characterization and operational interpretation .
specifically , it was proven in @xcite and very recently experimentally observed in @xcite that all nonclassical states of a finite - dimensional system can be turned into states with distillable entanglement between the system and a set of ancillae by an _
activation protocol_. focusing on a bipartite setting , the protocol runs as follows .
the subsystems @xmath1 and @xmath2 are first subject to arbitrary local unitary transformations @xmath3 ; then , each system @xmath4 interacts via a controlled - not ( cnot ) operation @xmath5 ( i.e. a so - called premeasurement interaction ) with an auxiliary system @xmath6 , @xmath4 , initialized in a pure state @xmath7.the activation protocol then possesses two key properties : i ) for all classical states @xmath0 at the input of the protocol , there exist local unitaries @xmath3 for which the output state @xmath8 is separable across the @xmath9 splitting , and ii ) for all nonclassical states @xmath0 and for all local unitaries , the output state is entangled across the @xmath9 splitting . let us stress that both criteria i ) and ii ) must be met by any scheme in order to be a valid activation protocol .
in particular , they allow us to define faithful measures of nonclassical correlations for the input state @xmath0 in terms of the output @xmath9 entanglement , minimized over @xmath3 .
one such measure , when the output entanglement is quantified by the negativity @xcite , has been termed negativity of quantumness @xcite , and has been experimentally investigated in @xcite in this paper we study activation of nonclassical correlations in multimode bipartite gaussian states @xmath0 of continuous variable systems @xcite .
nonclassical correlations of gaussian states have been studied extensively both theoretically and experimentally @xcite but their interplay with entanglement has not been pinned down so far in terms of the activation framework .
attempts to devise activation - like protocols for gaussian states have been explored @xcite . however , these differed significantly from the original prescription in that nonunitary operations were employed between system and ancillae , so that the entanglement generation was obtained as a dynamical feature , and conditions i ) and ii ) were not generally verified . here
we consider a general gaussian activation protocol in which @xmath3 are gaussian unitaries and the cnot gates are replaced with a global gaussian unitary on subsystems @xmath10 . in section [ sec_1 ]
we then prove that any such protocol satisfying condition i ) will unavoidably violate condition ii ) , which implies that activation of gaussian nonclassical correlations by gaussian operations is impossible .
this fact establishes a new no - go theorem for gaussian quantum information processing , which can be enlisted alongside other well known no - go results such as the the no - distillation theorem , according to which distilling entanglement from gaussian states by using only gaussian operations is impossible @xcite .
we then show in section [ sec_2 ] how , by using non - gaussian operations which properly extend the cnot to infinite dimensions , one can construct the continuous variable counterpart of the activation protocol of @xcite , verifying criteria i ) and ii ) .
this allows us to define the negativity of quantumness for gaussian states and to calculate it for relevant examples in section [ sec_3 ] .
this work provides an operational setting to understand and manipulate nonclassical correlations in paradigmatic infinite - dimensional systems .
we draw our conclusions in section [ sec_4 ] , while some technical derivations ( which can be of independent interest ) are deferred to the appendices .
gaussian states are quantum states of systems with an infinite - dimensional hilbert space ( continuous variable systems ) , e.g. a collection of harmonic oscillators , which possess a gaussian - shaped wigner function in phase space @xcite . @xmath11 modes are described by a vector @xmath12 of quadrature operators @xmath13 satisfying the canonical commutation rules expressible in terms of elements of the vector @xmath14 as @xmath15=i\omega_{jk}$ ] , @xmath16 with @xmath17 , where @xmath18 is the pauli @xmath19-matrix .
an @xmath11-mode gaussian state @xmath20 is fully characterized by a @xmath21 vector @xmath22 of the first moments with elements @xmath23 and by its @xmath24 covariance matrix ( cm ) @xmath25 with elements @xmath26 , @xmath27 , where @xmath28 .
gaussian unitaries are generated by hamiltonians that are quadratic in the quadrature operators and they preserve the gaussian characteristic of quantum states .
an @xmath11-mode gaussian unitary @xmath29 is represented in phase space by a @xmath24 real symplectic transformation @xmath30 satisfying the condition @xmath31 , which transforms a cm @xmath25 to @xmath32 .
here we are interested in the question of whether an activation protocol exists satisfying conditions i ) and ii ) which would rely solely on gaussian states and gaussian unitaries .
we therefore assume the state @xmath0 to be a gaussian state of @xmath33 modes with cm @xmath34 and the state @xmath35 of the ancilla to be also a gaussian state with cm @xmath36 .
the local unitaries @xmath3 of the original discrete protocol are replaced with local gaussian unitaries @xmath37 and @xmath38 represented by the symplectic matrices @xmath39 and @xmath40 , respectively .
likewise , the global operation @xmath41 on the whole system @xmath42 is replaced with one global gaussian unitary @xmath29 represented by a symplectic matrix @xmath30 .
let us recall the definition of a fully classical state @xcite .
suppose @xmath0 is a bipartite state containing two subsystems @xmath1 and @xmath2 with @xmath43 and @xmath44 modes respectively , and let @xmath45 be a basis of subsystem @xmath46 , with @xmath47 , @xmath48 and @xmath49 .
if there exists a basis @xmath50 consisting of the tensor products of all elements of @xmath51 with all elements of @xmath52 , then @xmath0 is a classical state if it is diagonal with respect to @xmath50 .
it has been shown in @xcite that a two - mode gaussian state is classical if and only if it is a product state , i.e. , its cm is represented by a direct sum @xmath53 of local cms @xmath54 .
one can prove that this statement remains valid for the generic case of bipartite @xmath33-mode gaussian states ( see appendix [ secapp_1 ] for the proof ) .
therefore , all non - product bipartite gaussian states ( including separable ones ) are nonclassical . according to condition
i ) in any gaussian activation protocol with an input gaussian product state there must exist local gaussian unitaries @xmath3 for which one gets a separable state @xmath8 across the @xmath9 splitting at the output of the protocol .
we will show however , that this implies that for _ all _ separable gaussian states including nonclassical ones , there exist local gaussian unitaries @xmath3 for which the output state is separable .
that is , condition ii ) is not satisfied .
thus , any gaussian activation protocol described above can not meet simultaneously criteria i ) and ii ) , and hence does not exist .
is a separable gaussian state prepared from a gaussian product state @xmath55 by correlated displacements @xmath56 and @xmath57 distributed according to a gaussian distribution with correlation matrix @xmath58 , eq .
( [ p ] ) . @xmath37 and @xmath38 are local gaussian unitaries which are adjusted such that without the displacements @xmath56 and @xmath57 the activation protocol produces from the product state @xmath55 an output state which is separable across the @xmath9 splitting .
as the unitaries @xmath37 , @xmath38 and @xmath29 induce a linear transformation of quadrature operators of the input modes , the displacements @xmath56 and @xmath57 can be relocated behind the global transformation @xmath29 ( dotted arrow ) .
the new displacements @xmath59 and
@xmath60 , eq . ( [ displacement ] ) , can not turn a separable state into an entangled state and therefore the protocol transforms the separable state @xmath0 into a state which is separable across the @xmath9 cut ( thick dashed line ) . see text for details.,width=321 ] the proof of this no - go theorem is depicted in fig .
it follows from the decomposability of any gaussian separable state into a product state and noise @xcite , and the linearity of symplectic transformations .
namely , for any separable gaussian state with cm @xmath34 there exist local cms @xmath54 such that @xmath61 in other words , any separable gaussian state with cm @xmath34 can be prepared from a suitable product state with cm @xmath53 by the addition of noise , represented by a positive - semidefinite matrix @xmath58 , i.e. , @xmath62 .
the noise can be created by displacing the vector of quadratures @xmath63 of the product state as @xmath64 . here
@xmath65 is a @xmath66 matrix given by the first @xmath67 columns of the matrix @xmath68 bringing the matrix @xmath58 to the diagonal form @xmath69 , where @xmath70 denote @xmath71 strictly positive eigenvalues of the matrix ( [ p ] ) , and @xmath72 is the vector of classical displacements uncorrelated with the vector of quadratures @xmath73 and distributed according to the gaussian distribution with zero means and the diagonal correlation matrix @xmath74 .
let us now consider a separable state with cm @xmath34 at the input of a gaussian activation protocol and let @xmath75 be a cm of the product state from which the state can be prepared using the aforementioned algorithm .
assume that the local symplectic matrices @xmath76 are chosen such that the cm @xmath77 of the output state , where @xmath36 is the cm of the state of the ancilla , is separable across the @xmath9 splitting .
hence , for the original separable state with cm @xmath34 , the output of the activation protocol is obtained by displacing the vector of quadratures @xmath78 for the state with cm @xmath79 by @xmath80 where @xmath81 is a @xmath82 zero vector with @xmath83 being the number of modes of the ancilla @xmath84 . however , for a separable state with cm @xmath79 , where @xmath85 is separable from @xmath84 , the local displacements ( [ displacement ] ) can not create a state in which the system @xmath85 is entangled with the system @xmath84 .
consequently , for any separable state ( even nonclassical ) it is always possible to find local gaussian unitaries for which the output is separable , thus accomplishing the proof of the no - go theorem .
therefore , gaussian operations are unable to activate nonclassical correlations of gaussian separable states into entanglement in the worst - case scenario : assuming condition i ) holds , then for any gaussian separable state there exist local gaussian unitaries for which the output of the activation protocol remains a separable gaussian state .
this indicates that a non - gaussian element , like a non - gaussian global unitary @xmath86 or a non - gaussian state of the ancilla , is necessary for faithful activation of nonclassical correlations in gaussian states . in the following we design such an activation protocol involving a non - gaussian cnot gate in the fock basis and an ancillary system in a gaussian state
the main benefit of the activation protocol is that it allows one to quantify the amount of nonclassical correlations in a given quantum state as the potential to create entanglement in the activation protocol @xcite .
more precisely , if @xmath87 denotes an entanglement measure quantifying the amount of entanglement between systems @xmath85 and @xmath84 in a quantum state @xmath8 , then we can define a measure of nonclassical correlations on the input state @xmath0 as @xmath88 where the minimization is carried out over all local unitaries @xmath89 and @xmath90 on subsystems @xmath1 and @xmath2 .
it has been proven in @xcite that @xmath91 , with equality if @xmath0 is pure .
from now on we assume that systems @xmath1 and @xmath2 each contain one mode .
the non - gaussian activation protocol is obtained as a direct generalization of the finite - dimensional protocol @xcite . at the input we allow for generally non - gaussian states @xmath0 of continuous variable systems , local unitaries @xmath92 and
@xmath93 , and the global gaussian unitary @xmath29 of the preceding protocol is replaced with the tensor product @xmath94 of the infinite - dimensional generalizations of cnot gates in the fock basis , @xmath95 where @xmath96 , @xmath97 , and @xmath98 is the @xmath99th fock state of mode @xmath100 .
we also assume the initial state @xmath35 of the ancilla @xmath84 to be the vacuum state @xmath101 .
hence , the final output state can be expressed as @xmath102 where @xmath103 by following arguments similar to the finite - dimensional case @xcite , one can show that the non - gaussian activation protocol defined above satisfies both criteria i ) and ii ) . for condition
i ) we assume that @xmath0 is classically correlated and hence there exist local unitaries @xmath92 and @xmath93 such that the density matrix @xmath104 , eq .
( [ tilderho ] ) , takes the form @xmath105 . making use of eqs .
( [ fockcnot ] ) and ( [ rhoout ] ) it then follows that the output state of the protocol is the following convex mixture of product states , @xmath106 , and is thus a separable state across the @xmath9 splitting as required . for the proof of condition
ii ) we now suppose that the density matrix @xmath0 is nonclassical and show that then the density matrix @xmath8 given in eq .
( [ rhoout ] ) is entangled across the @xmath9 cut for all local unitaries @xmath92 and @xmath93 . to prove the presence of entanglement in @xmath8 we will use the negativity @xmath107 defined in @xcite as @xmath108 here @xmath109 denotes the trace norm
, @xmath110 is the partial transpose @xcite of the state @xmath8 with respect to subsystem @xmath85 , and a strictly positive value of negativity implies that the state @xmath8 is ( distillable ) entangled with respect to the @xmath9 splitting .
the specific feature of the present activation protocol is that the output state @xmath8 is a so - called maximally correlated state and therefore , following results in @xcite , the output negativity can be expressed as @xmath111 where @xmath112 , @xmath113 and @xmath114 are elements of the density matrix @xmath104 , eq .
( [ tilderho ] ) , in the fock basis .
since our input state @xmath0 is nonclassical , the state @xmath104 is also nonclassical for any choice of unitaries @xmath92 and @xmath93 .
thus , there must be at least one non - zero off - diagonal element @xmath115 for every choice of @xmath92 and @xmath93 .
hence , eq . ( [ n2 ] ) implies @xmath116 and the output state @xmath8 is entangled for any nonclassical input state .
this completes the proof of our non - gaussian activation protocol .
the optimization in eq .
( [ qe ] ) is generally carried out over all local unitary operations @xmath89 and @xmath90 , including non - gaussian ones , which is not a tractable task .
here we consider input gaussian states with cm in standard form @xcite , and consider the non - optimized output entanglement @xmath87 obtained when the local unitaries @xmath3 are selected to be identity matrices .
therefore , the state ( [ tilderho ] ) remains a gaussian state in standard form with the following cm @xmath117 where @xmath118 , @xmath119 and @xmath120 are diagonal matrices . in what follows
we determine the non - optimized quantity for some classes of two - mode gaussian states by considering the negativity ( [ n1 ] ) as an entanglement measure @xmath121 , and using eq .
( [ n2 ] ) .
the corresponding measure of nonclassical correlations @xmath122 is called the negativity of quantumness @xcite accordingly .
although our choice of local unitaries @xmath3 gives in general an upper bound on @xmath123 , we find that it coincides with the true measure on pure states , leading us to conjecture that our choice is optimal for calculating the negativity of quantumness of all two - mode gaussian states in standard form .
verifying this conjecture numerically is beyond the scope of this work
. a closed form of the output negativity can be found for pure two - mode gaussian states .
the density matrix @xmath104 amounts to that of the two - mode squeezed vacuum state , with @xmath124(\tanh r)^{m_1+n_1}\delta_{m_1,m_2}\delta_{n_1,n_2}$ ] , where @xmath125 is the squeezing parameter .
hence , by a direct substitution into eq .
( [ n2 ] ) we get @xmath126 consequently , as the output negativity @xmath127 is equal to the negativity of the input state @xmath0 , it coincides with the true optimized negativity of quantumness @xmath122 @xcite , and our choice of local unitaries is thus optimal for pure states .
the negativity ( [ np ] ) is depicted by a solid red line in fig .
[ fig2 ] .
( [ np ] ) ] ( solid red line ) and its lower bound @xmath128 [ eq . ( [ lp ] ) ] ( dash - dotted brown line ) for pure squeezed vacuum states , plotted as a function of the local mean number of thermal photons @xmath129 .
upper bound on the negativity of quantumness @xmath130 [ eq .
( [ nm ] ) ] ( dashed blue line ) and its lower bound @xmath131 [ eq . ( [ lm ] ) ] ( dotted black line ) for separable mixed states obtained as unbiased mixtures of coherent states , plotted as a function of the local mean number of thermal photons @xmath132 .
the inset shows a close - up for @xmath133 , where the lower bounds become tight.,width=321 ] these gaussian states are of the form @xmath134 and can be prepared by splitting a thermal state with mean number of thermal photons @xmath135 on a balanced beam splitter . here
@xmath136 , @xmath137 and @xmath138 . the states are already in standard form with a cm ( [ tildegamma ] ) specified by @xmath139 and @xmath140 . making use of the components of a coherent state in fock basis
@xmath141 we get the following matrix elements of the state ( [ mixture ] ) , @xmath142 where @xmath143 . by substitution of the latter expression into eq .
( [ n2 ] ) we get after some algebra @xmath144 ^ 2 - 1\right\}.\ ] ] the negativity ( [ nm ] ) is depicted by a dashed blue line in fig .
[ fig2 ] , and is generally smaller than the one of pure states calculated in ( [ np ] ) .
both classes of gaussian states have a nonzero negativity of quantumness which increases with @xmath145 ; this is in agreement with earlier studies of nonclassical correlations based on entropic measures of quantum discord @xcite .
in general we need the fock basis elements @xmath115 for an arbitrary two - mode gaussian state with zero first moments . combining the results of refs .
@xcite we can express them as @xmath146 where @xmath147 is the cm of the state , @xmath148 is the @xmath149 identity matrix , and @xmath150 is the four - dimensional hermite polynomial @xcite at the origin ; see appendix [ secapp_2 ] for a complete derivation of eq .
( [ rhohermite ] ) . here
@xmath151o^{\dagger}v\ ] ] is the symmetric matrix defining the polynomial , where @xmath152 and @xmath153 for the standard - form cm @xmath147 , eq .
( [ tildegamma ] ) , we get in particular @xmath154 with @xmath155 being the @xmath156 identity matrix , @xmath157 and @xmath158 ( @xmath159 ) .
one can then evaluate the negativity ( [ n2 ] ) by performing a numerical summation of the absolute values of the elements ( [ rhohermite ] ) .
the higher - order hermite polynomials can be calculated from the lower - order ones by using e.g. the recurrence formula derived in appendix [ secapp_2 ] .
we remark that the compact expression in equation ( [ rhohermite ] ) is of independent interest and can be useful for the characterization of hybrid information processing involving conversion between continuous and discrete variable entanglement @xcite , or particularly for studies of bell nonlocality of arbitrary two - mode gaussian states by means of dichotomic pseudospin measurements @xcite , whose expectation value can be conveniently evaluated at the fock space level . in the context of the present paper , apart from the utility for numerical evaluation of the output negativity ( [ n2 ] ) , equation ( [ rhohermite ] ) also enables us to derive a simple analytical lower bound on the output negativity .
the bound results from the following chain of inequalities @xmath160 where the first inequality follows from the inequality @xmath161 which holds for any @xmath162 , the second inequality is a consequence of the triangular inequality for absolute values , and the last equation follows from the expression for the generating function of the four - dimensional hermite polynomials at the origin @xcite , @xmath163 where @xmath164 and @xmath165 is the matrix ( [ rr ] ) . a comparison between the right - hand side ( rhs ) of the previous equation and the expression of the husimi @xmath166-quasiprobability distribution @xmath167 in the fock basis further yields @xmath168 as can be easily seen from the results of appendix [ secapp_2 ] .
therefore , the last expression in the chain of inequalities ( [ inequalities ] ) can be written in the following compact form @xmath169 now , making use of the inequalities ( [ inequalities ] ) and equality ( [ phia11 ] ) one finds that the sum in ( [ n2 ] ) is lower - bounded as @xmath170 which finally gives the following bound on the output negativity ( [ n2 ] ) @xmath171.\ ] ] the bound ( [ l ] ) can be evaluated for any zero - mean two - mode gaussian state with cm @xmath147 by calculating the matrix ( [ rr ] ) and substituting it into the formula ( [ phia11 ] ) . to test the tightness of the bound
we calculate it for the previous examples of pure states and mixtures of coherent states , and compare the obtained lower bounds with the exact values of the negativities ( [ np ] ) and ( [ nm ] ) , respectively .
the cm @xmath147 is in the standard form ( [ tildegamma ] ) in both cases and therefore one can evaluate easily the matrix ( [ rr ] ) using eqs .
( [ rst ] ) and ( [ rj ] ) which gives , after substitution into eq .
( [ phia11 ] ) , @xmath172 for pure states , and @xmath173 for unbiased mixtures of coherent states .
the corresponding negativities then satisfy @xmath174\equiv\mathcal{l}_{\rm p}\ ] ] and @xmath175 the bounds @xmath128 and @xmath131 as well as the negativities @xmath176 , eq .
( [ np ] ) , and @xmath130 , eq .
( [ nm ] ) , are depicted in fig .
the figure shows that both bounds are tight in the region of small @xmath177 ( see the inset ) , which also proves that eq .
( [ nm ] ) amounts to the exact value of the negativity of quantumness for mixtures of coherent states with small mean number of thermal photons in each mode .
both lower bounds are then shown to increase with increasing @xmath178 and the gap between the bounds @xmath179 and the numerically evaluated values of the output negativities @xmath180 gets larger .
further analysis reveals however that the lower bounds @xmath128 and @xmath131 are nonmonotonic for larger @xmath177 ; they both attain a maximum at @xmath181 and @xmath182 , respectively , and then both monotonically decrease for larger values of @xmath178 ; eventually , both lower bounds become trivial as they enter the region of negative values , namely @xmath183 for @xmath184 and @xmath185 for @xmath186 . as a final remark , note that the sum in negativity ( [ n2 ] ) just amounts to the so - called @xmath187-norm of the density matrix @xmath104 @xcite , i.e. , @xmath188 .
the results of the present section thus also describe how to calculate numerically the @xmath187-norm for an arbitrary two - mode gaussian state with zero means and the inequality ( [ sumbound ] ) gives a simple analytical lower bound @xmath189 on such a norm .
we have shown that a protocol capable of activating nonclassical correlations in bipartite gaussian states based solely on gaussian operations can not exist .
we have also constructed a non - gaussian activation protocol and we have investigated quantitatively its performance using the negativity of quantumness as a figure of merit .
our analysis suggests that optimal performance of the protocol is achieved if the input gaussian state is in the standard form .
restricting to the local gaussian unitaries the conjecture can be proved or disproved with the help of eq .
( [ rhohermite ] ) by numerical minimization of the negativity ( [ n2 ] ) with respect to the unitaries , which is left for further research .
we believe that our results will stimulate further exploration of the negativity of quantumness and its interplay with other nonclassicality indicators @xcite in the context of gaussian states .
l. m. acknowledges the project no .
p205/12/0694 of gar and the european social fund and msmt under project no .
d. m. acknowledges the support of the operational program education for competitiveness project no .
cz.1.07/2.3.00/20.0060 co - financed by the european social fund and czech ministry of education .
g. a. acknowledges the brazilian agency capes [ pesquisador visitante especial - grant no .
108/2012 ] and the foundational questions institute [ grant no .
fqxi - rfp3 - 1317 ]
. g. a. would also like to thank m. barbieri and m. piani for discussions .
this section is dedicated to the proof that a bipartite gaussian state @xmath0 of an @xmath43-mode subsystem @xmath1 and an @xmath44-mode subsystem @xmath2 is classically correlated across the @xmath190 splitting if and only if it is a product state @xmath55 .
the proof of the `` only if '' part is trivial because any product state is diagonal in the product of eigenbases of local states .
the `` if '' part can be proved using the necessary and sufficient condition for zero quantum discord @xcite . quantum discord @xmath191 of a quantum state @xmath0 with a measurement on subsystem @xmath2 is zero if an only if the state can be expressed as @xcite @xmath192 where @xmath193 is an orthonormal basis of subsystem @xmath2 .
the zero - discord criterion @xcite then says that a quantum state @xmath0 can be expressed in the form ( [ qcstate ] ) if and only if for an informationally complete positive operator valued measurement ( ic - povm ) on subsystem @xmath1 , the conditional states @xmath194 of subsystem @xmath2 corresponding to the measurement outcomes @xmath99 , mutually commute , i.e. , @xmath195=0,\quad \mbox{for all $ k$ and $ k'$}.\ ] ] we consider a gaussian state @xmath0 with zero means and covariance matrix ( cm ) @xmath34
. modes @xmath196 comprising the subsystem @xmath1 are subject to a gaussian measurement characterized by a cm @xmath197 and a vector of measurement outcomes @xmath198 .
if a measurement outcome @xmath99 occurs then the state @xmath0 collapses into the @xmath44-mode state @xmath194 of subsystem @xmath2 with cm @xmath199 and vector of first moments @xmath200 of the form @xcite @xmath201 where @xmath202 and @xmath203 are blocks of the cm @xmath34 expressed with respect to the @xmath190 splitting , @xmath204 as in ref .
@xcite we will now express criterion ( [ criterion ] ) in terms of the characteristic function .
for this purpose we will first use the fact that an @xmath44-mode quantum state @xmath205 can be expressed as @xcite @xmath206 where @xmath207 is the characteristic function of the state @xmath205 and @xmath208 is the displacement operator with @xmath209 and @xmath210 is the vector of quadratures . due to the validity of the relation @xmath211=(2\pi)^{m}\delta(\xi-\xi')$ ] we get from eq .
( [ rhoj ] ) immediately the following expression for the characteristic function of the state @xmath205 : @xmath212.\ ] ] making use of eq .
( [ rhoj ] ) we can express the commutator on the left - hand side ( lhs ) of eq .
( [ criterion ] ) as @xmath213=\frac{1}{(2\pi)^{2m}}\int\int_{\mathbb{r}_{2m}}c_{k}(\xi)c_{k'}(\xi ' ) \left(e^{-\frac{i}{2}\xi^{t}\omega\xi'}-e^{\frac{i}{2}\xi^{t}\omega\xi'}\right)w^{\dag}(\xi+\xi')d\xi d\xi',\ ] ] where @xmath214 and @xmath215 are the characteristic functions of the states @xmath194 and @xmath216 , respectively , and where we have used the relation @xmath217 with @xmath218 from eqs .
( [ sigma ] ) and ( [ dk ] ) it follows that the gaussian states @xmath194 and @xmath216 possess the same cm @xmath199 and first moments @xmath200 and @xmath219 , respectively , and therefore their characteristic functions read @xmath220 equation ( [ commutator ] ) allows us to calculate the characteristic function of the commutator @xmath221 $ ] given by @xmath222w(\xi)\right\}.\ ] ] by inserting the rhs of the commutator from eq .
( [ commutator ] ) into eq .
( [ ckkprimed ] ) , using eq .
( [ wdagw ] ) and carrying out the integration , we arrive at the characteristic function ( [ ckkprimed ] ) in the form @xmath223.\ ] ] from eq .
( [ ckkprimed ] ) and the formula @xmath224=\frac{1}{(2\pi)^{m}}\int_{\mathbb{r}_{2m}}c_{kk'}(\xi)w^{\dag}(\xi)d\xi\ ] ] it follows that @xmath221=0 $ ] if and only if @xmath225 for all @xmath226 . assuming that the cm @xmath199 in eq .
( [ sigma ] ) has finite second moments and the measurement outcomes @xmath99 and @xmath227 and hence also the displacements @xmath200 and @xmath219 defined by eq .
( [ dk ] ) are finite , the condition @xmath225 for all @xmath226 is equivalent to the condition @xmath228 which can be rewritten using eq .
( [ dk ] ) as @xmath229 previous results allow us to rephrase the zero - discord criterion of ref .
@xcite for bipartite gaussian states and gaussian ic - povms as follows .
an @xmath230-mode gaussian state @xmath0 can be expressed in the form ( [ qcstate ] ) if and only if the condition ( [ condition2 ] ) is satisfied for all @xmath231 , where @xmath99 and @xmath227 are measurement outcomes of an gaussian ic - povm on subsystem @xmath1 characterized by the cm @xmath197 .
condition ( [ condition2 ] ) is satisfied for all @xmath231 ( @xmath232 ) if and only if the matrix @xmath233 consider now the heterodyne measurement which is an example of a gaussian ic - povm @xcite .
then @xmath234 , the matrix @xmath235 is invertible and therefore condition ( [ condition3 ] ) is equivalent with the equation @xmath236 .
as both the matrices @xmath237 and @xmath238 are also invertible the latter condition is equivalent with the condition @xmath239 . for the heterodyne detection
the condition ( [ condition3 ] ) is thus equivalent with the vanishment of the off - diagonal block @xmath203 given in eq .
( [ gammaabblock ] ) , which carries intermodal correlations .
this means in other words , that a bipartite @xmath33-mode gaussian state can be expressed in the form ( [ qcstate ] ) if and only if it is a product state , i.e. , @xmath240 .
let us now move to the necessary and sufficient condition for a bipartite @xmath33-mode gaussian state to be a classically correlated state .
a quantum state @xmath0 is classically correlated if and only if @xmath241 @xcite , where @xmath242 is the discord of @xmath0 for measurement on subsystem @xmath1 .
a quantum state @xmath0 is therefore classically correlated if and only if it can be expressed simultaneously in the form ( [ qcstate ] ) and in the form @xmath243 according to the criterion given in @xcite a quantum state @xmath0 can be expressed in the form ( [ cqstate ] ) if and only if for an ic - povm on subsystem @xmath2 the conditional states @xmath244 of subsystem @xmath1 corresponding to the measurement outcomes @xmath99 mutually commute , i.e. , @xmath245=0,\quad \mbox{for all $ k$ and $ k'$}.\ ] ] like in the previous case we can express the latter condition in terms of a characteristic function .
we can proceed exactly along the same lines as in the case of the commutator ( [ commutator ] ) with the only difference that now we consider measurement on the @xmath44-mode subsystem @xmath2 .
consequently , the formulas which we get for the present case of the commutator ( [ commutator2 ] ) are obtained from the formulas derived in the context of commutator ( [ commutator ] ) by the replacements @xmath246 , @xmath247 of the blocks of the matrix @xmath34 and by the replacement @xmath248 .
thus we find that the commutator ( [ commutator2 ] ) vanishes if and only if @xmath249 .
therefore , the condition @xmath239 is necessary and sufficient for an @xmath33-mode gaussian state to be classical , which concludes our proof .
our aim is to express the elements of a density matrix of a gaussian state @xmath20 of two modes @xmath1 and @xmath2 in the fock basis .
here and in what follows we assume that the state has all first moments equal to zero .
the present derivation combines the results obtained in refs .
firstly we express the elements of the density matrix in the basis of coherent states as @xmath250 where we have used the expression of the components of a coherent state @xmath251 in the fock basis @xmath252 the matrix element on the lhs of eq .
( [ cohelements ] ) can be further expressed as @xmath253 where @xmath254^{-1}\alpha}\ ] ] is the husimi @xmath166-quasiprobability distribution of the gaussian state @xmath20 @xcite . here
, @xmath255 and @xmath256 is the complex cm corresponding to antinormal ordering of the canonical operators .
substituting now from eq .
( [ lhs ] ) into the lhs of eq .
( [ cohelements ] ) and making use of eq .
( [ phia ] ) we arrive at the following equality @xmath257^{-1}-\openone\right\}\alpha}\\ & & \quad = \sum_{m_1,m_2,n_1,n_2=0}^{\infty}\frac{{\alpha_1^*}^{m_1}{\alpha_2^*}^{m_2}\alpha_1^{n_1}\alpha_2^{n_2}}{\sqrt{m_1!m_2!n_1!n_2!}}{\langlem_1m_2|}\rho{|n_1n_2\rangle}. \nonumber\end{aligned}\ ] ] the lhs of the latter equation can be expressed in terms of the multi - dimensional hermite polynomials @xcite . specifically , the generating function of the four - dimensional hermite polynomials is @xmath258 where @xmath164 , @xmath259 , and @xmath165 is a symmetric matrix of order four .
the lhs of eq .
( [ focksum ] ) then can be rewritten in terms of the lhs of eq .
( [ generatingfunction ] ) as follows .
the complex cm @xmath256 can be expressed as @xmath260 where @xmath148 is the @xmath149 identity matrix , @xmath261 is a @xmath149 unitary matrix , and @xmath25 is the standard real symmetrically ordered cm of the state @xmath20 , with elements @xmath262 , @xmath263 , where @xmath264 is the @xmath265-th component of the vector of quadratures @xmath266 .
hence we get @xmath267^{-1}-\openone = o\left[\left(\gamma+\frac{1}{2}\openone\right)^{-1}-\openone\right]o^{\dagger}.\ ] ] furthermore , we can write @xmath268 where @xmath269 consequently , @xmath270^{-1}-\openone\right\}\alpha = h^t rh,\ ] ] where @xmath271o^{\dagger}v.\ ] ] as @xmath272 and the cm @xmath25 is symmetric , one finds immediately that @xmath273 and therefore @xmath165 is symmetric as required . making use of eqs .
( [ focksum ] ) and ( [ generatingfunction ] ) we get @xmath274 where the matrix @xmath165 defining the hermite polynomial @xmath275 is given in eq . ( [ r ] ) . by equating each term in the summation
we are left with the elements of the density matrix @xmath20 in the fock basis , @xmath276 where @xmath277 equation ( [ rhofock ] ) allows us to calculate any element of a density matrix in the fock basis for an arbitrary two - mode gaussian state with zero first moments . to calculate matrix ( [ r ] )
it is convenient to express the cm @xmath25 in the block form @xmath278 this allows us to express the inverse matrix @xmath279 , appearing in eq .
( [ r ] ) , in block form using the following blockwise inversion formula @xcite , @xmath280 higher - order hermite polynomials can be calculated from lower - order polynomials using a recurrence relation .
it is derived from the generating function ( [ generatingfunction ] ) , where we set @xmath281 . by deriving both sides of the equation ( [ generatingfunction ] ) with respect to the @xmath265-th element of the vector @xmath164 , substituting the rhs of eq .
( [ generatingfunction ] ) for the exponential function @xmath282 appearing on the lhs of the obtained expression and equating each term in the summation , we arrive at the following recurrence relation @xmath283 where @xmath284 is the four - dimensional hermite polynomial at the origin with multi - index @xmath285 .
the coefficients @xmath286 correspond to the @xmath287-th element of the matrix @xmath165 , eq .
( [ r ] ) , and @xmath288 is the @xmath265-th canonical basis vector with 1 in the @xmath265-th component and zeros everywhere else . here ,
any hermite polynomial with a negative index is zero , i.e. @xmath289 for all @xmath290 with @xmath291 for some @xmath265 .
every hermite polynomial at the origin can be found from the latter recurrence formula and by using the first few cases , @xmath292 with @xmath293 .
these can be derived by a direct calculation from the expression @xmath294 found in @xcite .
note that it is sufficient to calculate only the polynomials where the parity of the multi - index @xmath290 is even .
when the parity of the multi - index @xmath290 is odd , i.e. @xmath295 , where @xmath296 , then @xmath297 .
g. adesso , v. dambrosio , e. nagali , m. piani , and f. sciarrino , phys .
* 112 * , 140501 ( 2014 ) .
g. vidal and r. f. werner , phys .
a * 65 * , 032314 ( 2002 ) .
t. nakano , m. piani , and g. adesso , phys .
a * 88 * 012117 ( 2013 ) .
i. a. silva _
et al . _ , phys . rev .
110 * , 140501 ( 2013 ) ; f. m. paula _ et al .
lett . * 111 * , 250401 ( 2013 ) .
g. adesso and f. illuminati , j. phys .
a : math . theor . * 40 * 7821 , ( 2007 ) ; c. weedbrook _ et al . _ rev .
phys . * 84 * , 621 ( 2012 ) ; g. adesso , s. ragy , and a. r. lee , open syst . inf . dyn . * 21 * , 1440001 ( 2014 ) .
l. s. madsen , a. berni , m. lassen , and u. l. andersen , phys .
lett . * 109 * , 030402 ( 2012 ) ; r. blandino _ et al .
lett . * 109 * , 180402 ( 2012 ) ; m. gu , _ et al .
_ , _ nature phys .
_ * 8 * , 671 ( 2012 ) .
s. rahimi - keshari , c. m. caves , and t. c. ralph , phys .
a * 87 * , 012119 ( 2013 ) .
r. f. werner and m. m. wolf , phys .
lett . * 86 * , 3658 ( 2001 ) .
a. peres , phys .
* 77 * , 1413 ( 1996 ) .
r. simon , phys .
lett . * 84 * , 2726 ( 2000 ) . v. v. dodonov , v. i. manko , and v. v. semjonov , nuovo cimento b * 83 * , 145 ( 1984 ) .
v. v. dodonov , o. v. manko , and v. i. manko , phys .
a * 50 * , 813 ( 1994 ) .
j. fiurek and j. peina , quantum statistics of light propagating in nonlinear optical couplers . in j. peina ,
editor , _ coherence and statistics of photons and atoms _ , chapter 2 , pages 65 - 110 .
j. wiley , new york , 2001 . , edited by a. erdlyi ( mcgraw - hill , new york , 1953 ) .
a. datta , eprint - arxiv:0807.4490 .
g. giedke and j. i. cirac , phys . rev .
a * 66 * , 032316 ( 2002 ) .
g. m. dariano , p. perinotti , and m. f. sacchi , j. opt .
b * 6 * , s487 ( 2004 ) .
b. bylicka and d. chruscciski , phys .
a * 81 * , 062102 ( 2010 ) .
j. peina , _ quantum statistics of linear and nonlinear optical phenomena _
( kluwer , dordrecht , 1991 ) .
r. a. horn and c. r. johnson , _ matrix analysis _ ( cambridge university press , cambridge , england , 1985 ) . | we study general quantum correlations of continuous variable gaussian states and their interplay with entanglement .
specifically , we investigate the existence of a quantum protocol activating all nonclassical correlations between the subsystems of an input bipartite continuous variable system , into output entanglement between the system and a set of ancillae . for input gaussian states
, we prove that such an activation protocol can not be accomplished with gaussian operations , as the latter are unable to create any output entanglement from an initial separable yet nonclassical state in a worst - case scenario .
we then construct a faithful non - gaussian activation protocol , encompassing infinite - dimensional generalizations of controlled - not gates to generate entanglement between system and ancillae , in direct analogy with the finite - dimensional case .
we finally calculate the negativity of quantumness , an operational measure of nonclassical correlations defined in terms of the performance of the activation protocol , for relevant classes of two - mode gaussian states . |
lattice distortions are abundant in frustrated magnets as it is an efficient way to relieve the frustration and lift the macroscopic degeneracy of the classical ground states .
this has been , for example , an obstacle to synthesize and study materials that can be described by the quantum heisenberg model with spin-1/2 moments on ideal " lattice structures with isotropic spin exchange couplings .
the interest on the ideal " models stems from various theoretical suggestions for novel emergent quantum paramagnetic phases such as quantum spin liquid@xcite and valence bond solid ( vbs)@xcite phases . on the other hand
, those deformations may lead to different kind of emergent quantum ground states and understanding precisely how these phases are connected to the quantum ground states in the ideal " limit would help us understand more global picture of the quantum frustrated magnets.@xcite one of the most studied frustrated magnets is the quantum spin-1/2 antiferromagnetic heisenberg model on the kagome lattice .
there exist several materials that are candidates for the realization of this system.@xcite it has been , however , found that many materials undergo lattice deformations and quite often this leads to magnetically ordered phases at sufficiently low temperatures.@xcite so far , there are a small number of examples of two - dimensional kagome or kagome - like materials with spin-1/2 moments that do not show any magnetic ordering down to very low temperatures.@xcite the herbertsmithite , zncu@xmath1(oh)@xmath6cl@xmath0,@xcite is known for the ideal " kagome lattice structure , but it suffers from inter - site mixing between cu@xmath3 and zn@xmath3 , and makes the interpretation of the susceptibility data difficult . in the volborthite , cu@xmath1v@xmath0o@xmath7(oh)@xmath0@xmath82h@xmath0o,@xcite the kagome structure is deformed in an orthorhombic fashion , leading to two inequivalent spin exchange couplings between cu@xmath3 moments . in both cases , the specific heat at low temperatures reveals the absence of ( or a very small ) excitation / spin gap ( down to 50mk in herbertsmithite , for example ) , leading to various suggestions for exotic quantum ground states.@xcite more recently , it is found that cu@xmath3 moments in rb@xmath0cu@xmath1snf@xmath2 form the layers of a deformed kagome lattice and do not show any magnetic order at low temperatures.@xcite the susceptibility data clearly shows a spin excitation gap ( in contrast to the previous two materials ) which is of the order of 20k while the average strength of the exchange coupling is estimated to be around 200k , which leads to the expectation that the ground state may be a vbs state .
cu@xmath1snf@xmath2 . here
all the exchange coupling constants are positive and satisfy @xmath9 . in this work , we consider the case of @xmath10 .
( b ) the j@xmath11-depleted kagome lattice that is topologically equivalent to the deformed kagome lattice when @xmath12 but @xmath13 .
, width=8 ] in this paper , we investigate the quantum ground states of the spin-1/2 antiferromagnetic heisenberg model on a series of deformed kagome lattices that may be relevant to rb@xmath0cu@xmath1snf@xmath2 .
each deformed kagome layer can be described by four possibly different nearest - neighbor exchange couplings , labeled by @xmath14 , @xmath15 , @xmath16 , and @xmath17 in decreasing order in magnitude as inferred from different cu@xmath3
cu@xmath3 bond angles ( see fig . [
fig : latticestructure](a ) ) . in rb@xmath0cu@xmath1snf@xmath2 ,
@xmath14 , @xmath15 , and @xmath16 are similar in magnitude while the magnitude of @xmath17 is about a half of the others.@xcite hence it is reasonable to start from a model where @xmath14 = @xmath15 = @xmath16 = @xmath19 . here
we study the evolution of the quantum ground states as a function of @xmath20 , where @xmath5 corresponds to the ideal kagome lattice .
using perturbative arguments starting from isolated dimer limits , we find that there exist two competing vbs phases on these lattices ; one with a 12-site unit cell ( vbs-12 , shown in fig.[fig : vbsstructure](a ) ) and another with a 36-site unit cell ( vbs-36 , shown in fig.[fig : vbsstructure](b ) ) .
the vbs-36 phase is smoothly connected to the vbs phase with a large unit cell on the ideal kagome lattice , which has been suggested as a strong contender for the quantum ground state of the spin-1/2 antiferromagnetic heisenberg model.@xcite the perturbation theory clearly shows that the vbs-12 is the lower energy ground state when @xmath21 , namely on the @xmath22-depleted lattice . on the other hand ,
the vbs-36 becomes the ground state when @xmath5 , consistent with the previous studies on the ideal kagome lattice .
we study the relative stability of these phases for an arbitrary deformation parameter @xmath23 using the perturbation theory , dimer series expansion , and bond operator mean field theory .
thus our work can also test the stability of the vbs phase with a 36-site unit cell on the ideal kagome lattice against the deformation from the ideal kagome lattice structure .
it is found that the vbs-36 phase on the ideal kagome lattice is quite sensitive to the deformation and only 3@xmath24 reduction in the magnitude of @xmath22 is enough to induce the instability of the vbs-36 state of the ideal kagome lattice .
thus we propose that the ground state for @xmath25 is the vbs-12 state .
note that the vbs-12 state is clearly the ground state of @xmath26-only model or when @xmath27 .
it is remarkable that the vbs-12 state is stable even when the strength of @xmath28 and @xmath29 are equal to @xmath26 .
thus we conclude that the vbs phase discovered in rb@xmath0cu@xmath1snf@xmath2 is most likely the vbs-12 state . in order to provide a future experimental test for the vbs-12 state ,
we compute the triplet dispersion spectra as shown in fig.[fig : dispersion ] .
the rest of the paper is organized as follows . in sec.[sec : candidate ] , we use perturbative arguments about isolated dimer limits to show that there exist two competing vbs phases - the vbs-12 and vbs-36 - on the @xmath22-depleted lattice . in sec.[sec : groundproperties ] , the dimer series expansion and the bond operator mean field theory are developed to investigate the relative stability of the two candidate vbs phases for an arbitrary strength of @xmath30 .
the triplet dispersion spectra is also discussed in detail for the vbs-12 state
. finally , in sec.[sec : discussion ] , we discuss the implications of our results to theory and experiments .
we can obtain valuable information about the spatial ordering pattern of singlet dimers by applying the perturbation theory around an isolated dimer limit.@xcite to perform the calculation , we divide the hamiltonian into two parts , _
i.e. _ @xmath31 . here
@xmath32 describes the heisenberg spin exchange interaction between two nearest - neighbor spins constituting a singlet dimer .
the inter - dimer interactions are contained in @xmath33 which is treated as a perturbation .
the ground state of the unperturbed hamiltonian @xmath32 , _
i.e. _ @xmath34 can be taken as the direct product of a set of isolated dimer singlets lying on the @xmath17-depleted kagome lattice .
let us start from two possible local configurations of two neighboring dimer singlets as shown in fig .
[ fig : secondorder](a ) and ( b ) .
note that a pair of neighboring dimers can be aligned in parallel , connected by a single unoccupied link ( fig .
[ fig : secondorder](a ) ) or they can be aligned in a perpendicular direction if one end of a dimer is simultaneously connected to both spins of its neighboring dimer ( fig .
[ fig : secondorder](b ) ) .
the important point is that the ground state energy can be lowered through second order perturbation processes only when the two dimers are lying in parallel .
when two dimers are lying in a perpendicular direction , @xmath33 @xmath34=0 because of the odd parity of the singlet state with respect to the reflection and there is no energy shift due to the second order processes . the energy gain from each configuration where two neighboring dimers are lying in parallel is @xmath35 . while a dotted line indicates the interdimer coupling contained in the perturbation , @xmath33 .
( a ) neighboring dimers are lying in parallel .
there are energy gains @xmath36 in the second order and @xmath37 in the third order .
( b ) neighboring dimers are lying in a perpendicular geometry .
there is no energy gain in the perturbation theory ( explained in the text ) . here
the triangle carrying a dimer is called a _ filled _ triangle .
( c ) an _ empty _ triangle .
there is no energy gain in the third order perturbation theory .
( d ) a _ perfect _ hexagon .
there is an additional energy gain @xmath38 in the third order perturbation theory .
, width=7 ] in order to determine the global structure of the vbs phases , however , we need to take into account all the geometric constraints for possible vbs patterns .
note that the basic building blocks of the @xmath17-depleted kagome lattice are triangles , hexagons and bridge links which connect each triangle to its neighboring hexagons ( shown in fig .
[ fig : latticestructure](b ) ) .
one can show that the number of triangles , @xmath39 , is given by @xmath40 , where @xmath41 is the number of spins ( or lattice sites ) on the lattice .
similarly , the number of hexagons , @xmath42 , and the number of bridge links , @xmath43 , satisfy @xmath44 and @xmath45 , respectively .
the total number of links , @xmath46 , on the lattice is then given by @xmath47 since the number of dimers , @xmath48 , is equal to @xmath49 and each dimer is lying on a single link , the total number of the unoccupied links is @xmath50 .
the unoccupied links connecting neighboring dimers in various local dimer configurations are shown as the dotted lines in fig .
[ fig : secondorder ] according to the result of the second order perturbation theory for two neighboring dimers , when one of the links of a triangle is occupied by a dimer singlet ( we call such a triangle as a _ filled _ triangle ) , the other two links would generate no energy gain from the second order processes because these two unoccupied links are simultaneously connected to one end of a nearby dimer , generating a ` perpendicular ' arrangement between the filled link of a triangle and a nearby dimer . hence if we denote the number of the _ filled _ triangles as @xmath51 , the number of the unoccupied links that connect two nearby dimers in parallel configuration is @xmath52 .
therefore , the total energy gain per spin from the second order processes is given by @xmath53 note that the condition @xmath51=0 uniquely determines the ground state vbs configuration which is shown in fig . [ fig : vbsstructure](a ) . here
the unit cell consists of six dimer singlets which lie in a pinwheel structure and we call this vbs state as the vbs-12 state . in the vbs-12
, every triangle is an _ empty _
triangle , that is , no link of a triangle is occupied by a dimer singlet .
this is the dimer configuration one would obtain when only @xmath26 is present or @xmath27 ( see fig .
[ fig : latticestructure](a ) ) .
it is interesting to note that the vbs-12 state is stable all the way to the limit where @xmath28 and @xmath29 become comparable to @xmath26 .
-depleted lattice .
the light dotted ( yellow ) lines on the lefthand - side pictures show the @xmath22 links .
( a ) vbs-12 : a valence bond solid with a 12-site unit cell .
( b ) vbs-36 : a valence bond solid with a 36-site unit cell . here
the thick dotted ( pink ) lines indicate the unit cells .
, width=8 ] now we extend the perturbation theory to the third order in @xmath54 .
let us first consider two neighboring dimers lying in parallel as shown in fig.[fig : secondorder](a ) .
in addition to the energy gain from the second order perturbation , @xmath55 , this local configuration allows a third order process which results in the energy gain , @xmath56 . on the other hand , the dimer configuration in fig.[fig : secondorder](b )
does not allow any third order process .
in fact , this configuration is locally an eigenstate of the hamiltonian and does not produce any virtual state in any order of the perturbation theory .
similarly , the unoccupied links in fig.[fig : secondorder](c ) making an _ empty _ triangle , would not generate any energy gain in the third order , either .
therefore when we calculate the third order energy gain coming from various local configurations , we have to remember that the unoccupied links on every _ filled _ and _ empty _ triangles would not produce any third order energy shift .
fig.[fig : secondorder](d ) shows another configuration in which a new third order process can be generated .
this structure , which is commonly known as a _ perfect _ hexagon , consists of three neighboring dimers lying on a hexagon . through
the resonating third order processes around a _ perfect _ hexagon , the ground state energy can be lowered by @xmath57 .
based on these results , the total energy gain per spin by the third order processes can be summarized as follows .
@xmath58 where @xmath59 represents the number of perfect hexagons and we use the fact that the number of triangles , @xmath60 , satisfies @xmath61 .
therefore if we can maximize the number of the filled triangles and perfect hexagons , we can get the largest energy gain from the third order processes .
the vbs-36 state depicted in fig.[fig : vbsstructure](b ) satisfies this requirement , that is , @xmath62 and @xmath63 .
note that the three neighboring hexagons lying in a triangular structure can not be ` perfect ' at the same time for any kind of dimer coverings .
interestingly , this vbs-36 phase is adiabatically connected to the 36-site - unit - cell vbs ground state of the ideal kagome lattice @xcite in the sense that as we increase the magnitude of @xmath64 from zero to one , the vbs-36 phase evolves smoothly to the 36-site - unit - cell vbs ground state of the ideal kagome lattice . in the case of the @xmath22-depleted lattice , the vbs ground state is already uniquely determined by the second order perturbation theory
that is , the vbs-12 phase is the ground state and the vbs-36 phase is a closely competing phase .
however , it is not clear which phase is more stable when @xmath17 has finite magnitude . in the case of the ideal kagome lattice
, all the dimer coverings are degenerate up to the second order in @xmath54 and the degeneracy can only be lifted by the third order perturbation processes as emphasized in ref .. this is because the numbers of the filled and empty triangles on the ideal kagome lattice are fixed for any dimer covering due to the spatial geometry of the ideal kagome lattice ( see ref . ) .
this is , in fact , the reason why the 36-site - unit - cell vbs is the most stable vbs state on the ideal kagome lattice .
therefore , we may expect that the change in magnitude of @xmath17 would affect the relative importance between the second order and the third order processes . in the next section ,
we investigate the effect of finite @xmath17 on the ground state properties of the deformed kagome lattice .
in order to refine the ground state energy estimation beyond the third order perturbation theory and confirm the results of the previous lowest order expansions , we develop a systematic scheme to compute the perturbation series using the linked cluster expansion method . in this approach
, the ground state energy can be expressed as the sum of the contributions from a sequence of finite clusters @xcite . to carry out the series expansion ,
all the occupied links that make up the dimer covering are given an interaction strength @xmath65 and all the other unoccupied links are given a strength @xmath66 .
we apply the series expansion method to compute the energy of the two competing vbs phases on the @xmath22-depleted lattice , @xmath67 , and the ideal kagome lattice limit , @xmath68 , up to the fifth order in @xmath54 .
the ground state energy per site of an infinite lattice can be written as a sum over cluster terms of the form , @xmath69 . here
@xmath70 is the so - called lattice constant for the cluster @xmath71 , which is the number of embeddings of the cluster @xmath71 per lattice site .
@xmath72 is the reduced energy of the cluster @xmath71 , which is defined recursively as @xmath73 . here
@xmath74 is the ground state energy of the cluster @xmath71 and the sum is over all subclusters @xmath75 of @xmath71 . since the reduced energy @xmath72 is equal to the sum of the contributions from all connected diagrams ( _ i.e. _ the patterns of connected dimers ) that span the cluster @xmath71 , such diagrams only contribute beyond a certain minimum order in @xmath54 , proportional to the size of the cluster
thus , to obtain a perturbation series that is exact up to a given order , it is sufficient to sum up the reduced energies of all the clusters up to the corresponding size .
this gives us a systematic procedure for carrying out the expansion to higher orders.@xcite another appealing aspect of applying the linked cluster expansion method is that we can considerably reduce the complexity of the calculation by using the freedom in choosing clusters .
@xcite in a recent work of the dimer series expansion study on the ideal kagome lattice@xcite , the unoccupied links on each triangle are grouped together , which results in significant simplification of the computation and rapid convergence of the series for the ground state energy . in this scheme , for any dimer covering on the ideal kagome lattice , every _
empty _ and _ filled _ triangles constitute the elementary clusters .
all the other clusters can be constructed by putting the elementary clusters together appropriately .
based on this specially designed clusters , it was shown that the series for the ground state energy can be determined up to the fifth order in @xmath54 using only five different graphs ( or diagrams).@xcite summing up the terms in the series , the ground state energy per spin of the vbs-36 phase on the ideal kagome lattice is estimated to be @xmath76 . for comparison , we compute the ground state energy of the vbs-12 phase on the ideal kagome lattice .
the same graphs can be used for the calculation and the summation of the terms in the series up to the fifth order in @xmath54 leads to the energy of @xmath77 . note that even though the vbs phase with a 36-site unit cell is the ground state of the ideal kagome lattice , the energy difference between the two vbs phases is quite small . -depleted kagome lattice that contribute to the ground state energy up to the fifth order in @xmath54 .
the thick dotted ( pink ) lines are the weak links on the hexagons while the thin dotted ( black ) links represent the weak links on the empty triangles . , width=8 ] applying similar strategies to the @xmath17-depleted kagome lattice , we compute the series for the ground state energy of the two competing vbs phases . we first consider the vbs-12 phase on the @xmath17-depleted kagome lattice . here
we group the three weak links on an empty triangle together and distinguish them from the other weak links on the hexagons . in this scheme ,
only five different graphs ( shown in fig.[fig : vbs12 - 4th ] ) contribute to the ground state energy up to the fifth order in @xmath54 .
the smallest cluster in fig.[fig : vbs12 - 4th](a ) is made of a single dimer and has the reduced energy , @xmath78 .
fig.[fig : vbs12 - 4th](b ) and ( c ) show two clusters whose contributions start from the second order in @xmath54 . the reduced energies of the corresponding clusters , _
i.e. _ , @xmath79 and @xmath80 respectively , can be obtained after subtracting the subcluster energy @xmath81 appropriately and are found to be @xmath82 finally , the resonance processes coming from the remaining two graphs in fig.[fig : vbs12 - 4th](d ) and ( e ) can lower the ground state energy through the fourth order processes .
the corresponding reduced energies of the two clusters , _
i.e. _ , @xmath83 and @xmath84 can be obtained by the same procedure used earlier and using the fact that the subclusters are made of the graphs in fig.[fig : vbs12 - 4th ] ( a ) , ( b ) and ( c ) .
since the reduced energies of all the other clusters that are larger in size than those in fig.[fig : vbs12 - 4th](d ) and ( e ) , begin with o(@xmath85 ) terms , the ground state energy can be determined exactly at least up to the fifth order in @xmath54 with those five graphs shown in fig.[fig : vbs12 - 4th ] .
after we determine the lattice constants of all the clusters , the ground state energy for the infinite lattice is found to be @xmath86 this leads to the estimation of the ground state energy , @xmath87 for the vbs-12 phase .
following similar procedures , the ground state energy of the vbs-36 phase is obtained as @xmath88 this leads to the estimation of the ground state energy , @xmath89 for the vbs-36 phase , which is higher than that of the vbs-12 state .
this is consistent with the previous results of the perturbation theory and the vbs-12 phase is the ground state of the @xmath17-depleted kagome lattice . in summary ,
the vbs-12 phase is lower in energy on the @xmath17-depleted kagome lattice ( @xmath13 ) while the vbs-36 phase has lower energy on the ideal kagome lattice ( @xmath90 ) .
based on this result , we expect that there should be an energy level crossing between these two limits as the magnitude of @xmath64 is changed from zero to one . in order to obtain the detailed information about the level crossing , we use the self - consistent bond operator mean field approach to the two phases for an arbitrary strength of @xmath30 in the next section .
we study the relative stability of the vbs-12 and vbs-36 phases using the bond operator mean field theory @xcite for the deformed kagome lattice with an arbitrary strength of @xmath91 . in this formulation ,
the dimer singlet degrees of freedom are used as natural building blocks and the quantum corrections coming from the triplet fluctuations can systematically be investigated @xcite .
( @xmath65 is set to one ) obtained from the bond operator mean field theory .
note that there is a level crossing around @xmath92 as indicated by an arrow .
, width=8 ] let us consider the two @xmath93 spins constituting a dimer singlet , @xmath94 and @xmath95 .
the hilbert space is spanned by four states that can be taken as a singlet state , @xmath96 , and three triplet states , @xmath97 , @xmath98 and @xmath99 .
then , the singlet and triplet boson operators are introduced such that each of the above states can be created from the vacuum @xmath100 as follows : @xmath101 to eliminate unphysical states from the enlarged hilbert space , the following constraint needs to be imposed on the bond - particle hilbert space : @xmath102 where @xmath103 and @xmath104 , and we adopt the summation convention for the repeated indices hereafter unless mentioned otherwise . constrained by this equation
, the exact expressions for the spin operators can be written in terms of the bond operators : @xmath105 where @xmath106 is the third - rank totally antisymmetric tensor with @xmath107 .
we consider the antiferromagnetic heisenberg model hamiltonian , @xmath108 , where the exchange couplings of the neighboring spins are taken to be @xmath109 ( see fig .
[ fig : latticestructure ] ) . utilizing the bond operator representation of spin operators ,
the hamiltonian can be rewritten solely in terms of bond particle operators .
the hard - core constraint among bond particle operators is imposed via the lagrange multiplier method .
the quartic interactions between the triplets in the resulting hamiltonian are treated by the self - consistent hartree - fock approximation .
the computational procedure for the vbs-36 phase is basically in parallel to that of the recent work on the 36-site - unit - cell vbs ground state of the ideal kagome lattice @xcite .
here we present some details of the bond operator theory applied to the vbs-12 phase on the deformed kagome lattice .
since the six dimers within the unit cell of the vbs-12 phase are equivalent as is obvious from the six - fold rotational symmetry of the ordering pattern in fig.[fig : vbsstructure](a ) , the singlet condensate density @xmath110 and the chemical potential @xmath111 can be set to be @xmath110 = @xmath112 and @xmath111=@xmath113 in our mean field theory .
here @xmath114 denotes the position of the unit cell and @xmath115 indicates the dimer index inside the unit cell .
the hard - core constraint on the bond - particle operators is imposed by adding the following lagrange multiplier term , @xmath116 .
the quartic interactions between the triplet particles are decoupled using the mean field order parameters @xmath117 and @xmath118 , where @xmath119 and @xmath120 .
we consider four mean - field variables , @xmath117 and @xmath118 , labeled by @xmath121 = 1 or 2 . here
@xmath122 and @xmath123 denote the diagonal and off - diagonal triplet correlations between the dimers residing in the _ same _ unit cell while the other two variables , @xmath124 and @xmath125 represent the triplet correlations between neighboring dimers in _ different _ unit cells .
the above four order parameters @xmath122 , @xmath123 , @xmath124 , and @xmath125 together with @xmath112 and @xmath113 are determined self - consistently by solving the coupled saddle point equations .
@xcite ( @xmath65 is set to one ) .
note that the lower flat mode is fourfold degenerate while the upper one is doubly degenerate when @xmath126 .
, width=8 ] in fig .
[ fig : groundenergy ] , we plot the ground state energies of the vbs-12 and vbs-36 phases as a function of @xmath127 .
when @xmath13 , the ground state energy of the vbs-12 ( vbs-36 ) from the bond operator mean field theory is @xmath128 ( @xmath129 ) which is close to the results from the series expansion study , @xmath130 ( @xmath131 ) .
as one increases the magnitude of @xmath17 , the ground state energy of the vbs-12 grows more rapidly than that of the vbs-36 phase and finally an energy level crossing occurs around @xmath92 . beyond this point
the vbs-36 phase becomes more stable and remains as the ground state up to the ideal kagome lattice limit .
the evolution of the triplet excitation spectrum for the vbs-12 phase shows interesting behavior as the magnitude of @xmath17 changes . in fig .
[ fig : dispersion ] , we plot the triplet dispersion spectra when @xmath132 and 1 , respectively .
note that the triplet dispersion of the vbs-12 phase on the ideal kagome lattice ( @xmath133 ) shows completely flat spectrum .
flat dispersion in the momentum space indicates the existence of localized eigenstates.@xcite in this case , the three neighboring dimers around an empty triangle support the localized eigenmodes . when @xmath134 , the six dimers within a unit cell are all ` orthogonal ' to each other in the sense that each spin of a dimer is simultaneously connected to the two spins of its neighboring dimers as shown in fig .
[ fig : vbsstructure](a ) .
in this situation , the triplet particles can fluctuate only through the dimers of its neighboring unit cell connected via empty triangles , and hence they are confined within that space .
the three localized triplet modes coming from the three dimers around an empty triangle consist of doubly degenerate e modes and a nondegenerate a mode ( in the standard convention of irreducible representations in group theory.@xcite ) .
since the unit cell consists of six dimers , each e mode and a mode is doubly degenerate . in fig .
[ fig : dispersion](c ) , the lower flat mode has fourfold degeneracy while the upper one is doubly degenerate .
the fact that all the triplet dispersions are completely flat is an interesting characteristic of the triplet spectrum of the vbs-12 phase on the ideal kagome lattice ( note , however , that the true ground state is the vbs-36 phase on the ideal kagome lattice ) . according to a recent work @xcite ,
the triplet spectra of the vbs-36 phase on the ideal kagome lattice contain some dispersive modes even though the number of flat modes is quite large there as well . in particular ,
the lowest flat band is degenerate with the second lowest dispersive band only at the zone center . as discussed in detail in ref .
, the degeneracy at a particular momentum point reflects the existence of noncontractible loop states and is topologically protected .
therefore , the completely flat spectrum of the vbs-12 phase on the ideal kagome lattice means that all the flat modes are topologically trivial .
this results from the fact that the size of every localized mode is not larger than the size of the unit cell and each localized eigenmode around an empty triangle is linearly independent .
the condition for the emergence of the topologically nontrivial structure in dispersions is discussed in detail in ref .. ( see also ref .. ) on the other hand , when the magnitude of @xmath17 is reduced , the degenerate bands split and triplet particles spread over the whole lattice . in the case of @xmath135 that may directly be relevant to rb@xmath0cu@xmath1snf@xmath2 ,
one can still see the gap between the lower four bands and the upper two bands .
however , when @xmath17 becomes vanishingly small , the gap structure disappears due to wide bandwidth of the triplet bands .
-depleted kagome lattice . here
the dotted ( red ) square represents the smallest flippable loop .
, width=8 ] in the case of the vbs phases with a big unit cell , a large number of degenerate dimer configurations may allow many singlet excitations below a spin gap .
for example , the vbs-36 order on the ideal kagome lattice leads to the large degeneracy that comes from the combination of the broken translational symmetry and the local resonance moves around flippable loops ( see ref . and ref . ) . here
the flippable loops mean a collection of alternating occupied and unoccupied links ; when the roles of the occupied and unoccupied links on a flippable loop are interchanged , one hardcore dimer covering can be continuously transformed to another one .
the perfect hexagons and the pinwheel dimers form the smallest flippable loops in the case of the 36-site vbs phase on the ideal kagome lattice.@xcite the number of singlet excitations arising from such considerations is still smaller than what was observed in the exact diagonalization studies @xcite of small system sizes ( up to 36 sites ) for the spin-1/2 antiferromagnetic heisenberg model on the ideal kagome lattice .
it was , however , argued that the triplets can scatter to form additional singlet bound states in finite size systems and this may explain the exact diagonalization results @xcite .
on the other hand , the vbs-12 phase on the @xmath17-depleted kagome lattice does not break any translational symmetry and the local resonances from loop flips are the only sources of low energy singlet excitations . in fig.[fig : loopflip ] , we show the smallest flippable loop consisting of the four occupied and four unoccupied links , and its loop flip in the vbs-12 phase .
note that there is a change in the number of the _ empty _ and _ filled _ triangles after the loop flips .
in fact , a single loop flip increases the number of the _ filled _ triangles by two . according to eq.([eq : e - secondorder ] ) and eq.([eq : e - thirdorder ] ) in the perturbation theory , this requires the energy cost of @xmath136 , which is much larger than the measured spin gap @xmath137 if the experimentally estimated @xmath138 is used .
since the number of the _ filled _ triangles involved in a loop flip is proportional to the size of the flippable loop , the singlet excitations coming from a larger loop flip require even higher energy cost .
therefore , the vbs-12 phase is quite stable against local singlet fluctuations and it seems difficult to produce the singlet wave " whose existence is speculated in a recent experimental paper .
@xcite we expect the lowest singlet excitation would be well above the spin gap , which can be contrasted with many singlet excitations below a spin gap in the case of the ideal kagome lattice . in summary , slight lattice distortions can completely change the nature of the ground state of the quantum antiferromagnet on the kagome lattice and this leads to the abrupt modification of the elementary excitation spectra in both the triplet and singlet channels . even though @xmath26 , @xmath28 , and @xmath29 are not exactly the same in rb@xmath0cu@xmath1snf@xmath2 , the magnitudes of these exchange couplings are very similar and satisfy @xmath139 .
interestingly , the vbs-12 phase studied in this work corresponds to the dimer covering that resides only on the @xmath26 links .
this suggests that the vbs-12 phase on the @xmath26 links is stable all the way to the limit where @xmath28 and @xmath29 become the same as @xmath26 .
thus , we expect that our results are directly applicable to rb@xmath0cu@xmath1snf@xmath2 .
in particular , our predictions on the triplet and singlet spectra can be tested in the future neutron scattering and specific heat experiments on rb@xmath0cu@xmath1snf@xmath2 .
we thank tyler dodds , michael j. lawler , and erik s. srensen for helpful discussions .
this work was supported by the nserc of canada , the canada research chair , and the canadian institute for advanced research . | motivated by a recent experiment on rb@xmath0cu@xmath1snf@xmath2 , where spin-1/2 cu@xmath3 moments reside on the layers of kagome - like lattices , we investigate quantum ground states of the antiferromagnetic heisenberg model on a series of deformed kagome lattices .
the deformation is characterized by a weaker exchange coupling @xmath4 on certain lattice links appropriate for rb@xmath0cu@xmath1snf@xmath2 with @xmath5 corresponding to the ideal kagome lattice . in particular , we study possible valence bond solid phases using the perturbation theory around isolated dimer limits , dimer series expansion , and self - consistent bond operator mean field theory . it is shown that the valence bond solid phase with a 36-site unit cell of the ideal kagome lattice is quite sensitive to a small lattice distortion as the kind discovered in rb@xmath0cu@xmath1snf@xmath2 . as a result , we find that a more likely quantum ground state in rb@xmath0cu@xmath1snf@xmath2 is the valence bond solid phase with a 12-site unit cell , where six dimers form a pinwheel structure , leading to strong modification of the elementary triplet and singlet excitation spectra in the deformed kagome lattices . |
the nand tree quantum walk is described by the transmission coefficient of a series of coupled states , shown in fig .
each `` state '' @xmath7 in the graph corresponds to the lowest orbital state of a quantum dot in the coulomb blockade regime , with @xmath8 labeling the bottom quantum dot at level @xmath9 and @xmath10 labeling the @xmath11th dot at the @xmath12th level .
we assume a large orbital level spacing @xmath13 , where @xmath4 is the average tunnel coupling in the tree of dots .
the coupled dot hamiltonian can be written : @xmath14 where @xmath15 describes the addition of a ( spin - less ) electron on the @xmath16th dot near the charge transition from @xmath17 .
each dot can have a slight detuning from the fermi energy of @xmath18 , and each dot @xmath16 is tunnel coupled to dots @xmath19 and @xmath20 with a tunnel coupling @xmath21 , forming a physical binary tree structure .
inputs in the top of the tree .
nodes are labeled from 1 to 15 .
each node can be realized as a quantum dot , with the links between nodes represented by tunnel coupling between linked dots .
detunings @xmath22 applied to dots @xmath23 at the `` top '' of the tree set the local potential to be off of the coulomb blockade peak for dot @xmath23 if @xmath24 .
the conductance of the dot @xmath25 at the bottom of the tree , near zero bias ( @xmath26 ) , encodes the result of the nand operation .
[ fig1],width=288 ] at the top of the tree is an oracle ( @xmath27 ) , a quantum system with electron number operators @xmath28 coupled to the binary tree via @xmath29 .
unlike previous work , we focus on a coupling that may be easy to implement physically : for example , the oracle could be a mesoscopic device capacitively coupled to the top of the tree
. we will proceed with the case of an oracle in an eigenstate of the @xmath30 operators with eigenvalues @xmath31 each taking values 0 or 1 .
this sets the detuning of dots at the top of the tree to @xmath32 .
the algorithm progresses by the injection of an electron wavepacket into the bottom of the tree .
this probes the green s function of the bottom quantum dot , which depends recursively on the green s functions for the left and right subtrees . for wavepackets with small energy spreads @xmath33 , the successive , coherent scattering events
up and down the tree lead to either total transmission or total reflection of the wavepacket , where the result ( transmission or reflection ) is the final `` answer '' to the nand query for an input specified by the bits @xmath22 .
a voltage probe in the linear response regime provides such an energy probe for the system : the conductance for electrons passing through the bottom of the tree encodes the result of the tree . to show that our approach works in the presence of dephasing and disorder , we develop a recursive description of the scattering green s function in the quantum dot tree in the spirit of previous work @xcite .
we now describe an approach to understanding the nand tree evaluation using a recursively defined green s function for single - particle propagation through the tree .
if we consider a `` tree '' segment with a single state , @xmath34 , at the bottom , we can define the projectors @xmath35,@xmath36,@xmath37 , where @xmath36 projects into the subspace associated with the left side of the tree ( all states connected to dot 2 ) and @xmath37 similarly projects into the subspace associated with the right side .
we can write the inverse green s function exactly @xcite : @xmath38 this gives a recursive relation for the projected green s function @xmath39 for a state @xmath16 in terms of the green s functions for the subtrees connected to states @xmath40 and @xmath41 .
the term @xmath42 is a measure of the charge - based dephasing the electron may be expected to encounter . at the top ( largest fanout ) of the tree , we can write the single - state green s function @xmath43 which truncates the recursion relation . we remark that this recursion is only successful because the tree contains no loops ; otherwise , eq . ( [ e : recursion ] ) would fail . in the disorder - free case ,
@xmath44 for all @xmath16 and @xmath45 except for dots at the top of the tree with input bits @xmath46 , as described above .
setting energy to units of @xmath4 , we consider the green s function at the top of the tree for the three different inputs .
we set @xmath47 to be the sign of the detuning to illuminate the role the alternating sign @xmath48 plays .
we assume @xmath49 . for dot @xmath11 ,
tunnel coupled to @xmath50 and @xmath51 , we get ( to @xmath52 ) : @xmath53 as the @xmath47 alternate such that each paired `` 1 '' input has @xmath54 , the third term has no energy shift .
we now identify green s functions of the forms @xmath55 where the `` 0 '' and `` 1 '' indicate the expected nand results for the inputs .
any dependence on @xmath56 is removed by the alternation of @xmath47 for paired `` 1 '' inputs .
we can include effect of disorder in the system , both in variations in @xmath57 and in variations in @xmath18 .
both are likely to occur due differences in fabrication , low frequency noise on electrostatic gates ( such as those used to tune the system to the multi - dot regime ) and charge - fluctuators in the system .
for the three different inputs ( 00,01,11 ) at the top of the tree : @xmath58 ^ 2 } \nonumber \\ & & \ \
+ \frac{1}{\frac{t_l^2}{\epsilon_l - i \gamma } + \frac{t_r^2}{\epsilon_r - ig } } \\
1,0 & \rightarrow & - e \frac{1}{t_r^2 } + \frac{\epsilon_r - i \gamma}{t_r^2 } \\ 1,1 & \rightarrow & \frac{1}{e + ( i \gamma + ( s_l t_l^2 + s_r t_r^2)/\delta - \epsilon_j)}\end{aligned}\ ] ] these have the same form as the disorder free results of ref .
so long as @xmath59 , where @xmath4 is now the minimum tunnel coupling .
more generally , we will assume @xmath60 and @xmath18 to be drawn from a normal distribution with means @xmath61 and standard deviations @xmath62 .
we can remove any explicit @xmath56 dependence as long as @xmath63 .
we now show that two subtree s described by green s functions of the above type lead to the appropriate type green s function at the joining of the two subtrees , recursively implementing the nand tree .
consider the general map for two subtrees green s functions @xmath64 and @xmath65 , parameterized by @xmath66 .
@xmath67 for the cases 00 , 10 , and 11 are given to @xmath68 : @xmath69 \\
-(\alpha_l e + i \gamma \beta_l ) , \frac{1}{\alpha_r e + i \gamma \beta_r } & \rightarrow & -\left [ \frac{\alpha_r}{t_r^2 } e + i \gamma \frac{\beta_r}{t_r^2 } \right ] \\
-(\alpha_l e + i \gamma \beta_l ) , -(\alpha_r e + i \gamma \beta_r ) & \rightarrow & \frac{1}{(1 + t_l^2 \alpha_l +
t_r^2 \alpha_r)e + i \gamma ( 1 + t_l^2 \beta_l + t_r^2 \beta_r + i \epsilon_j / \gamma)}\end{aligned}\ ] ] the overall map implements a nand gate by identification of `` 0 '' and `` 1 '' via eq .
[ e : logical ] . for a tree
consisting entirely of the `` most dangerous case '' ( an input of the form 1011 ) subtrees of ref . , we find @xmath70 , and @xmath71 , where the sum is over the set of all dots where both subtrees evaluate to 1 .
the first term indicates that the green s function derived is only appropriate within an energy range @xmath72 about @xmath26 .
this leads to the condition that the electron wavepacket be spread over a time @xmath73 , and sets @xmath74 scaling of the evaluation time . for random , uncorrelated noise , the second term ( with @xmath5 terms )
has a root - mean - square of @xmath75 , and leads to an energy shift by that amount from the desired resonance at @xmath26 as well as a dephasing @xmath76 .
this indicates that @xmath77 form the requirements for operation of the device .
less difficult is the bounding requirement @xmath78 .
we compared these results with numerical simulations and find consistent behavior for tree sizes up to @xmath79 . a sample with @xmath80 is shown in fig .
[ f : qdconduct ] .
we now detail how a quantum dot coupled to a linear voltage probe can measure the result of the nand tree well below the temperature limit of the leads .
this is a generalization of sub - temperature spectroscopy used in double quantum dot transport measurements @xcite .
we write the green s function , including the self energies @xmath81for left and right leads coupled to the dot via @xmath82 , as @xmath83 by working well above the band - bottom of the leads ,
we can write @xmath84 with @xmath85 , where @xmath86 is the density of states near the fermi energy . the meir formalism @xcite for the transmission @xmath87 $ ] yields @xmath88 - \gamma/2)^2 + ( e - t_1 ^ 2 { \rm re}[g_1(e ) ] - \epsilon_0)^2 } \ ] ] the conductance follows from the landauer formula , @xmath89 , where @xmath90 is the fermi distribution .
a few remarks are in order .
first , a pole at @xmath26 for @xmath91 leads to zero conductance , while a `` 1''-like green s function ( proportional to energy ) , leads to a peak at @xmath92 , thus letting the existence or absence of transport indicate a `` 1 '' or `` 0 '' result .
second , when dephasing ( @xmath93 ) is weak , the transmission function is a lorentzian with a width at most @xmath94 . when convolved with the derivative of the fermi function at @xmath95 ,
the edges of the fermi function are effectively damped , leading to sub - temperature energy resolution ( limited only by @xmath96 ) .
to see these effects , we simulate the expected conductance as a function of @xmath97 and @xmath98 at @xmath99 in fig . [
f : qdconduct ] for an input of @xmath80 bits .
cases resulting in `` 0 '' and `` 1 '' are both shown .
for this scenario , we set @xmath100 and @xmath101 .
the temperature dependence can be mitigated by further reductions of @xmath96 .
however , disorder , as described below , may limit the achievable resolution . with outputs `` 0 '' ( blue , purple ) and `` 1 '' ( red , orange ) .
the dependence of conductance on @xmath97 ( left ) and @xmath98 ( right ) are shown .
disorder parameters @xmath102 ( red , blue ) and @xmath103 ( orange , purple ) are used . [
f : qdconduct],width=288 ]
one difficulty encountered for making large nand trees using this approach is the increasing perimeter of the tree as a function of tree depth ( @xmath1 versus @xmath104 ) . for sufficiently large trees , the spacing between successive levels @xmath12 and @xmath105 of the tree must grow as @xmath106 for a @xmath107-dimensional layout .
furthermore , tunnel coupling exponentially decreases as the spacing between quantum dots increases , potentially limiting this approach to small trees . as an alternative
, `` dummy '' levels of the tree can be inserted : these are states coupled only to a single dot going up or down the tree ( fig .
[ f : hfractal]a ) .
each acts as an inverter , where the `` missing '' state is equivalent to the absence of a resonance in one `` leg '' of the tree , which is the `` 1''-type green s function .
specifically , a state of either `` 0 '' or `` 1 '' type parameterized by @xmath108 becomes , after coupling through two inverters , @xmath109 . going a distance @xmath107 ( through @xmath110 inverters ) gives @xmath111 as an explicit construction , we consider the h - fractal ( fig .
[ f : hfractal]b ) for the layout of quantum dots .
the h - fractal is a binary tree structure comprised of an `` h '' with the two sides the same length as the middle bar . , and an area - to - dot size ratio of @xmath112 for an @xmath113-level tree .
] this has the benefit that the distance between successive levels of the binary tree doubles for every _ two _ levels . a `` worst case '' trees ( again , with inputs 1011 ) has the mapping @xmath114 where the distance goes as @xmath115 at the @xmath12th level ( with @xmath113 the bottom of the tree ) .
a tree comprised entirely of worst - case trees has @xmath116 this indicates that a 2d architecture leads to an overhead of @xmath117 for the evaluation of the tree . this still performs better than the best classical algorithm . as a final remark
, we can selectively remove one of the pair of dots to produce a `` not '' operation inline with the following nand operation by opening the tunnel coupling between the pair of dots .
this allows the tree evalutor to evaluate _
boolean functions , not just nand trees .
taking values @xmath118ev , @xmath119ev , @xmath120 mev , @xmath121ev , @xmath122ev , and @xmath123ev ( 20 mk dilution refrigerator electron temperature ) , we find the limiting algorithm size @xmath124 , where the limit is due to the disorder in the system .
using the h - fractal layout , this corresponds to a mesoscopic region of area @xmath125mm@xmath126 for dot center - to - dot center spacing @xmath127 nm .
this includes appropriate space for electrostatic gates or etched barriers .
phase coherent transport over such large regions have been demonstrated in a number of experiments @xcite .
the expected evaluation time is of order @xmath128 ns ( to generate a differential signal of ten electrons , sufficient for fast detection ) .
we can use this nand tree evaluator as a subsystem in a larger processing unit .
the best known classical algorithm evaluates in @xmath129 time . using the quantum device as a sub - tree evaluator of a quantum oracle , a problem of size @xmath130
can be evaluated in @xmath131 time , as subtrees of size @xmath132 can be calculated sequentially using the quantum subsystem as a hardware accelerator .
better hybrid quantum - classical approaches may exist , but require further investigation .
this paper illustrates a fundamental and heretofore unrecognized quality of the quantum nand algorithm .
in particular , the entire nand quantum walk can be done with a single - electron wavepacket and appropriate resonances ( quantum dots ) .
other approaches using classical waves , such as light in coupled cavities , would in principle have the same green s function .
thus , from a quantum information perspective , the interesting part of the algorithm lies in the oracle : the quantum mechanical system which `` sets '' the inputs .
we heretofore considered the oracle to be in an eigenstate of the number operators @xmath133 .
if we describe the boolean function evaluation of the tree by @xmath134 or 1 , an arbitrary oracle state leads to an expected result of the tree of @xmath135 .
for example , a two - level tree evaluates @xmath136 , a non - trival , high - order correlation function of the oracle .
we now have the outstanding task of identifying interesting quantum systems to use as an oracle .
the author thanks e. farhi , j. goldstone , s. gutmann , c. marcus , and s. jordan , and the workshop on solid state quantum information systems at the niels bohr international academy .
this research is supported by the pappalardo fellowship . | we propose a physical implementation of the quantum nand tree evaluation algorithm .
our approach , based on continuous time quantum walks , uses the wave interference of a single electron in a heirarchical set of tunnel coupled quantum dots .
we find that the query complexity of the nand tree evaluation does not suffer strongly from disorder and dephasing , nor is it directly limited by temperature or restricted dimensionality for 2-d structures .
finally , we suggest a potential application of this algorithm to the efficient determination of high - order correlation functions of complex quantum systems . recently a new quantum algorithm was discovered that evaluates a binary tree of incommeasurate conditions via the logical nand gate in a time faster than the provably best classical algorithm @xcite .
one construction of this algorithm relies upon a continuous time quantum walk between coupled bound states , where the `` answer '' to the posed query is determined by the reflection or transmission of a single - particle wave - packet through an idealized , complex scattering device @xcite .
mesoscopic transport in semiconductor heterostructures have ideal properties for creating a device to implement a continuous time quantum walk .
their long phase coherence times and lengths for electronic transport allows for novel studies of the electronic properties of small - scale structures @xcite .
modern fabrication techniques allow for complex structures with robust controls and fast manipulation @xcite , and the key features of the continuous time quantum walk approach ( a fano - like intereference effect with a non - trivial set of coupled bound states ) have been experimentally demonstrated @xcite .
in this paper we propose an approach to implementing nand trees by mimicking the logical structure of the hamiltonian formulation with a physical arrangement of quantum dots . using a recursive green s function approach inspired by ref . , we find the quantum speed - up of the quantum nand algorithm that it takes a time @xmath0 for @xmath1 input bits is maintained if the phase decoherence rate @xmath2 @xcite and the disorder ( detunings and tunnel couplings ) be lower than @xmath3 , where @xmath4 is the characteristic strength of interdot tunnel coupling @xmath4 and @xmath1 is the number of bits in the input to the nand tree . one key feature of the present work is the use of a large physical space ( @xmath5 quantum dots are used ) to allow for efficient coupling to the complex quantum system ( the `` oracle '' ) to which the query is posed .
we develop appropriate techniques such that temperature and restricted dimensionality lead to at most a @xmath6 overhead in the evaluation of the algorithm .
we conclude with a discussion of the `` quantum - ness '' of the algorithm and consider how such a device can be used to measure many - body properties of complex quantum systems . |
the topological and dynamical aspects of complex networks have been the focus of intensive research during the last years @xcite .
an open and unsolved problem in network and computer science is the following question : how to cover a network with the fewest possible number of boxes of a given size @xcite ? in a complex network , a box size can be defined in terms of the chemical distance , @xmath1 , which corresponds to the number of edges on the shortest path between two nodes .
this means that every node is less than @xmath1 edges away from another node in the same box .
here we use the burning approach for the box covering problem @xcite , thus the boxes are defined for a central node or edge . instead of calculating the distance between every pair of nodes in a box , the maximal distance to the central node or edge @xmath2
this distance can then be related to the size of the box @xmath3 for a central node and @xmath4 for a central edge .
the maximal chemical distance within a box of a given size @xmath5 is @xmath6 for a central node and @xmath7 for a central edge .
although this problem can be simply stated , its solution is known to be np - hard @xcite .
it can be also mapped to a graph coloring problem in computer science @xcite and has important applications , e.g. , the calculation of fractal dimensions of complex networks @xcite or the identification of the most influential spreaders in networks @xcite . here
we introduce an efficient algorithm for fractal networks which is capable to determine the minimum number of boxes for a given parameter @xmath1 or @xmath2 .
moreover , we compare it for two benchmark networks with a standard algorithm used to approximately obtain the minimal number of boxes . in principle , the optimal solution should be identified by testing exhaustively all possible solutions . nevertheless , for practical purposes , this approach is unfeasible , since the solution space with its @xmath8 solutions is too large .
present algorithms like maximum - excluded - mass - burning @xcite and merging algorithms @xcite are based on the sequential addition of the box with the highest score , e.g. , the score is proportional to the number of covered nodes , and the boxes with the highest score are sequentially included .
other algorithms are based on simulated annealing @xcite , but without the guarantee of finding the optimal solution .
even greedy algorithms end up with a similar number of boxes as the algorithms mentioned before @xcite .
the greedy algorithm sequentially includes a node to a present box , if all other nodes in this box are within the chemical distance @xmath1 and if there is no such box , a new box with the new node is created .
it is therefore believed that the results are close to the optimal result , although the real optimal solution is unknown .
+ this paper is organized as follows . in section
ii , we introduce the algorithm and then explain the main difference between the present state of the art algorithm and our optimal algorithm for a given distance @xmath2 . in section iii , results for two benchmark networks are presented and the improvement in performance of our algorithm is quantitatively shown . finally , in section iv , we present conclusions and perspectives for future work .
we use two slightly different algorithms for the calculation of the optimal box covering solution , one for odd values of @xmath1 and another for even values @xmath1 . to get the results for an odd value ,
the following rules are applied : + 1 . create all possible boxes : for every node @xmath9 create a box @xmath10 containing all nodes that are at most @xmath11 edges away .
node @xmath9 is called center of the box . an example is shown in fig . [ fig:1]a .
2 . remove unnecessary boxes : search and remove all boxes @xmath10 which are fully contained in another box @xmath12 ( see fig . [
fig:1]b ) .
3 . remove unnecessary nodes : for every node @xmath9 , check all the boxes containing @xmath9 : @xmath13 .
if another node @xmath14 is contained in all of these boxes , remove it from _ all _ boxes ( see fig .
[ fig:1]c ) .
4 . remove pairs of unnecessary twin boxes : find two nodes @xmath15 which are both in exactly two boxes of size two : @xmath16 , @xmath17 and @xmath18 , @xmath19 . if @xmath20 and @xmath21 , then @xmath22 and @xmath23 can be removed . if @xmath24 and @xmath25 , then @xmath22 and @xmath26 can be removed .
an example for this rule is shown in fig .
[ fig:5 ] .
note that such twin boxes also appear for @xmath27 due to the removal of unnecessary nodes .
search for boxes that must be contained in the solution : add all boxes @xmath10 to the solution , which have a node @xmath9 only present in this box .
remove all nodes @xmath14 covered by @xmath10 from other boxes .
iterate a : repeat 2 - 5 until there is no node which is covered by a single box and is not part of the solution .
system split : identify if the remaining network can be divided into subnetworks , such that all boxes in a subnetwork contain only nodes of this subnetwork .
then these subnetworks can be processed independent from each other .
system split : find the node which is in the smallest number of boxes @xmath28 , each of these boxes covers another set of nodes @xmath10 .
if there is more than one node fulfilling this criterion , chose the node which is covered by the largest boxes .
then the algorithm is divided into @xmath28 sub - algorithms , which can be independently calculated in parallel . by removing from each of the @xmath28 sub - algorithm another set of nodes @xmath10 ,
all possible solutions are considered .
an example for the splitting is shown in fig .
[ fig:6 ] .
since we want to identify only one optimal solution , we do not need to calculate the results of all sub - algorithms .
as soon as one of the sub - algorithms identifies an optimal solution , we can skip the calculation of the others . furthermore , the calculation of a sub - algorithm can be skipped , if the minimal number of required additional boxes reaches the number of the , so far , best solution of a parallel sub - algorithm .
iterate b : repeat 2 - 8 until no nodes are uncovered .
identify the best solution : chose the solution with the lowest number of boxes .
this solution is optimal for a given @xmath2 .
+ to get the results for an even value of @xmath1 the first step is slightly different : 1 . create all possible boxes : for every _ edge _ @xmath9 create a box @xmath10 containing all nodes that are at most @xmath29 _ nodes _ away .
_ edge _ @xmath9 is called center of the box .
all other steps are the same as for the odd case .
note that the calculation for odd values scales with the number of nodes of the network @xmath30 and with the number of edges @xmath31 for even values .
instead of sequentially including boxes , the idea of our algorithm is to remove all non - optimal boxes from the solution space ending up with a final , optimal solution . to reduce the huge solution space ,
our box covering algorithm uses two basic ingredients : 1 ) unnecessary boxes from the solution space are discarded and the boxes which definitively belong to the solution are kept .
2 ) unnecessary nodes from the network are discarded .
these two steps reduce the solution space of a wide range of network types significantly , specially if they are applied in alternation as the removal of a box can lead to the removal of nodes and other boxes and vice - versa .
nevertheless these two steps do not necessarily lead to the optimal solution , thus the solution space has to be split into several possible sub - solution spaces . in each of these sub - solutions
the first two steps are repeated .
note that the splitting does not reduce the number of possible solutions , thus only the first two steps reduce the solution space and in the worst case , the algorithm must calculate the entire solution space .
in any case , for many complex networks iterating these three steps significantly reduces the solution space to a few solutions from which the optimal box covering can be obtained .
+ the remaining question is how to judge whether a box or node is necessary or unnecessary . on the one hand
a box is unnecessary if all nodes of a box are also part of another box .
this box can be removed , because the other box covers at least the same nodes and often additional nodes . on the other hand
a box is necessary if a node is exclusively covered by this single box .
this box has to belong to the solution , since only if the box is part of the solution , the node is covered .
+ in contrast , nodes can easily be identified as unnecessary .
for example all nodes of a box , which is part of the solution , can be removed from all other boxes , since they are already covered .
additionally , if a node shares all boxes with another node , the other node can be removed , since the second node is always covered , if the first node is covered .
these few rules are in principle sufficient to get the optimal solution , since our algorithm starts with _ all _ @xmath8 or @xmath32 ( for central edges ) possible solutions and discards unnecessary and includes necessary boxes .
+ although we only calculate results for undirected , unweighted networks , the algorithm can easily be extended to directed and weighted networks . in both cases
only the initial step , the creation of boxes , is different . for directed networks
, the box around a central node contains all nodes which are reachable with respect to the direction , while for weighted networks , the distance is the sum of the edge weights between the nodes .
+ next we show that our algorithm can also identify optimal solutions for large networks .
therefore , we have applied it to two different benchmark networks , namely the _ e. coli _ network @xcite , with 2859 proteins and 6890 interactions between them , and the www network @xcite .
we compare the results for the minimal box number @xmath33 of our algorithm for different values of box sizes @xmath1 with the results of the greedy graph coloring algorithm * ? ? ?
* , as displayed in fig .
[ fig : cellular ] .
while the absolute improvement is rather small , the relative improvement is up to @xmath34 larger for @xmath35 .
if the network is fractal , it should obey the relation , @xmath36 where @xmath37 is the fractal dimension .
interestingly , it seems that the fractal dimension @xmath38 from the greedy algorithm and @xmath39 from our optimal algorithm of the network is nearly unaffected by the choice of the algorithm . note that for @xmath40 , due to the fact that the boxes are calculated based on the definition of a central node or edge , we have one more box .
the simplest case where such difference occurs is in a chain of four connecting nodes ( 1 - 2 , 2 - 3 , 3 - 4 , 4 - 1 ) .
all nodes have the chemical distances of two to each other ( @xmath41 ) , however it is not possible to draw a box around a node with radius one ( @xmath42 ) , which contains all nodes .
+ the second example is the www network , containing 325729 nodes and 1090108 edges .
as in the previous case , our algorithm outperforms the state of the art algorithm , but yields similar fractal behavior , as shown in fig .
[ fig : www ] . for intermediate box sizes
@xmath43 , we have a large improvement since up to @xmath0 and up to @xmath44 fewer boxes are needed .
for @xmath45 we have two box more , like in the _ e. coli _ network case due to the two definitions of the box covering problem , while for larger @xmath1 both algorithm give similar results .
interestingly , it seems that the improvement for even distances @xmath1 ( for central edges ) is significantly larger than for odd distances @xmath1 ( for central nodes ) .
[ fig : www1 ] we show the influence of the sequence of adding nodes to the boxes on the results of the greedy algorithm . while the results of fig .
[ fig : www ] are the minimal values obtained from 50 independent starting sequences , we calculated 1500 realizations for a single box size @xmath46 .
the difference between the improvement is with @xmath47 and @xmath48 rather small .
the gap between the optimal solution and the greedy algorithm is too large , thus for practical purposes , the greedy algorithm will never find the optimal solution for this box size .
+ the results for these two benchmark networks demonstrate that our algorithm is more effective than the state of the art algorithms .
nevertheless , due to the rapid decay of the number of boxes for larger box sizes , the fractal dimension of the two benchmark networks is only slightly different when using the optimal box - covering algorithm in comparison with other algorithms .
in closing , we have presented a box - covering algorithm , which outperforms the known previous ones .
we have also compared our algorithm with the state of the art methods for different benchmark networks and detected substantial improvements .
moreover the obtained solutions are optimal as a result of the algorithm design , if the box size is defined as the maximal distance @xmath2 to the central node or edge .
for example , our approach can be useful for designing optimal commercial distribution networks , where the shops are the nodes , the storage facilities the box centers and the radius is related to the boundary conditions , like transportation cost or time .
we acknowledge financial support from the eth competence center coping with crises in complex socio - economic systems ( ccss ) through eth research grant ch1 - 01 - 08 - 2 and by the swiss national science foundation under contract 200021 126853 .
we also thank the brazilian agencies cnpq , capes , funcap and the inst - sc for financial support .
18 d. watts and s. strogatz , _ nature ( london)_*393 * 440 ( 1998 ) .
r. albert , h. jeong and a .-
barabsi , _ nature _ * 401 * 130 ( 1999 ) . m. bartheemy and l. a. n. amaral , _ phys .
rev . lett._*82 * 5180 ( 1999 ) .
a.l . lloyd and r.m .
may , _ science _ * 292 * , 1316 - 1317 ( 2001 ) .
r. cohen , s. havlin , and d. ben - avraham , _ phys .
rev . lett._*91 * 247901 ( 2003 ) .
m. barthelemy , a. barrat , r. pastor - satorras , and a. vespignani , _ phys .
lett . _ * 92 * , 178701 ( 2004 ) .
gonzlez , p.g .
lind , h.j .
herrmann , _ phys .
rev . lett._*96 * 088702 ( 2006 ) .
l.k . gallos , c. song , s. havlin , h.a .
makse , _ proc .
acad . sci._*104 * 7746 ( 2007 ) .
moreira , j.s .
andrade jr . , h.j . herrmann and j.o .
indekeu , _ phys .
rev . lett._*102 * 018701 ( 2009 ) .
h. hooyberghs , b. van schaeybroeck , a.a .
moreira , j.s .
andrade jr . , h.j . herrmann and j.o .
indekeu , _ phys .
rev . e_*81 * 011102 ( 2010 ) .
g. li , s.d.s .
reis , a.a .
moreira , s. havlin , h.e .
stanley and j.s .
andrade jr . , _ phys .
rev . lett._*104 * 018701 ( 2010 ) . h.j .
herrmann , c.m .
schneider , a.a .
moreira , j.s .
andrade jr . and
s. havlin , _ j. stat .
* p01027 * ( 2011 ) .
schneider , a.a .
moreira , j.s .
andrade jr .
, s. havlin and h.j .
herrmann , _ proc .
acad . sci._*108 * 3838 ( 2011 ) .
schneider , t. mihaljev , s. havlin , h.j .
herrmann , _ phys .
rev . e_*84 * 061911 ( 2011 ) .
a. vespignani , _ nature physics_*8 * 32 ( 2012 ) .
peitgen , h. jrgens and d. saupe , chaos and fractals : new frontiers of science ( springer)(1993 ) .
j. feder , fractals ( plenum press ) ( 1988 ) .
a. bunde and s. havlin ( eds . ) , fractals in science ( berlin : springer - verlag ) ( 1995 ) .
jensen and b. toft ( eds . ) , graph coloring problems ( new york : wiley - interscience ) ( 1995 ) t.h .
cormen , c.e .
leiserson , r.l .
rivest and c. stein , introduction to algorithms ( mit press ) ( 2001 ) .
c. song , s. havlin and h.a .
makse , _ nature _ * 433 * 392 - 395 ( 2005 ) .
c. song , l.k .
gallos , s. havlin and h.a .
makse , _ j. stat .
mech . _ * 03006 * ( 2007 ) .
garey and d.s .
johnson , computers and intractability ; a guide to the theory of np - completeness ( new york : w.h .
freeman ) ( 1979 ) s.h .
yook , f. radicchi and h. meyer - ortmanns,_phys .
e _ * 72 * 045105 ( 2005 ) .
g. palla , i. dernyi , i. farkas and t. vicsek _ nature _ * 435 * 814 ( 2005 ) .
zhao , h.j .
yang and b. wang , _ phys .
rev . e_*72 * 046119 ( 2005 ) .
c. song , s. havlin and h.a .
makse , _ nature physics _ * 2 * 275 - 281 ( 2006 ) .
goh , g. salvi , b. kahng and d. kim , _ phys .
lett . _ * 96 * 018701 ( 2006 ) .
a.a . moreira , d.r .
paula , r.n .
costa filho and j.s .
andrade jr . , _ phys .
e _ * 73 * 065101 ( 2006 ) .
m. kitsak , l. k. gallos , s. havlin , f. liljeros , l. muchnik , h. e. stanley , h.a .
makse , _ nature physics _
* 6 * 888 - 893 ( 2010 ) .
m. locci , g. concas , r. tonelli and i. turna , _ wseas trans .
_ * 7 * 371 - 380 ( 2010 ) .
zhou , z.q .
jiang and d. sornette , _ physica a _ * 375 * 741 - 752 ( 2007 ) . http://lev.ccny.cuny.edu/@xmath49hmakse/methods/methods.html . | the self - similarity of complex networks is typically investigated through computational algorithms the primary task of which is to cover the structure with a minimal number of boxes . here
we introduce a box - covering algorithm that not only outperforms previous ones , but also finds optimal solutions .
for the two benchmark cases tested , namely , the _ e. coli _ and the www networks , our results show that the improvement can be rather substantial , reaching up to @xmath0 in the case of the www network . |
spin precession measurements on atoms , molecules or solid state samples are susceptible to noise from magnetic field fluctuations . for measurements performed close to the surface of conductors ,
the magnetic johnson noise caused by thermal fluctuation of electrons in the conductor @xcite can contribute significantly to the uncertainty of the measurement .
this effect is a known challenge in atom chip experiments @xcite and precision magnetometry @xcite .
magnetic johnson noise is also becoming a challenge for experiments that measure the electron s electric dipole moment ( edm ) @xcite .
measurements of the electron edm are typically carried out in neutral atoms or molecules @xcite using some form of electron spin precession .
limitations associated with magnetic noise in edm experiments can be avoided by choosing neutral system with a small magnetic moment and/or a large sensitivity to the size of edm @xcite . even using a favorable system
, magnetic johnson noise must be minimised through careful choice of the materials placed close to the atoms or molecules .
this paper describes a simple , low cost set of electric field plates which will be used for an improved measurement of the electron edm using the ybf molecule @xcite .
the ybf experiment measures the phase difference between zeeman sublevels of the molecule in an applied electric field @xmath1 ( which defines the @xmath2-axis ) and an applied magnetic field @xmath3 .
the phase shift @xmath4 of a level has an electric part @xmath5 , where @xmath6 is the electron s edm and @xmath7 is the effective electric field interacting with the edm , and @xmath8 is the duration of the interaction .
there is also a magnetic part @xmath9 , where @xmath10 is the magnetic moment ( in our case the bohr magneton ) and @xmath11 is the magnetic field component parallel to @xmath1 .
details of the experiment are given in @xcite .
the uncertainty in @xmath6 should ideally be limited by the shot noise associated with the number of ybf molecules detected , but @xmath4 also has noise due to the fluctuations of @xmath11 . in order to benefit from current improvements that lengthen @xmath8 and increase the number of detected molecules
, it is now becoming important to reduce the magnetic noise , which is dominated by the johnson noise due to current fluctuations in the electric field plates @xcite .
let the time - dependent noise in @xmath11 be @xmath12 . we relate this to the corresponding spectral density of the noise power , @xmath13 ,
by the wiener - khintchine relation @xmath14 the rms phase uncertainty after interaction time @xmath8 is @xmath15 as noted in @xcite , we can use eq .
( [ eq : wktheorem ] ) to rewrite eq .
( [ eq : phaseerreq ] ) in the form @xmath16 eqation ( 38 ) of varpula and poutanen @xcite provides a formula for @xmath17 at distance @xmath2 from an infinite slab of conducting material at temperature @xmath18 , having thickness @xmath19 and resistivity @xmath20 . in reality
the molecule passes between two electric field plates , each contributing to the magnetic field noise , so we multiply eq.(38 ) of @xcite by a factor of @xmath21 : @xmath22 where @xmath23 is a complicated expression given in eq.(39 ) of @xcite .
the @xmath24 variable encodes the frequency dependence of the magnetic field noise via fourier integrals . the last generation of the ybf eedm experiment @xcite used aluminium electric field plates ( @xmath25 m ) of thickness @xmath26 mm .
the gap between the plates was 12 mm , so the distance from the molecules to the surface was @xmath27 mm . at that distance , the magnetic noise spectrum given by eq.([eq : bnz ] ) is plotted as the top ( blue ) line in fig .
[ fig : bnzplot ] .
the noise tends to a constant at low frequency , given by @xcite @xmath28 the integrand in eq.([eq : mungerd ] ) also involves the function @xmath29 , shown ( green , middle ) in fig .
[ fig : bnzplot ] . here
@xmath30 ms as that was the interaction time in the experiment .
after squaring and multiplying these two functions , numerical integration of eq.([eq : mungerd ] ) gives the phase noise as @xmath31 mrad .
this causes an uncertainty in the electron edm of @xmath32 e.cm for a single measurement over coherence time @xmath33 ms in an electric field of 8.3 kv / cm .
averaging over a data run of @xmath34 measurements the corresponding johnson noise in @xmath6 is @xmath35 e.cm . while this was negligible in the previous version of the experiment , that is no longer the case @xcite ,
so we have designed new plates with a much lower johnson noise
. quantities to be integrated in eq.([eq : mungerd ] ) .
blue : magnetic field noise spectrum @xmath36 at a temperature of @xmath37 k , calculated using eq .
[ eq : bnz ] between aluminium plates ( @xmath38 m ) of thickness @xmath39 cm and separation @xmath40 mm .
green : the function @xmath41 , taking an interaction time @xmath8 of 1 ms .
red ( dashed ) : @xmath36 between gold film plates ( @xmath42 m ) of @xmath43 nm thickness . ]
equation ( [ eq : bnzf0 ] ) suggests four ways to reduce the magnetic noise : reduce @xmath18 , increase @xmath2 , increase @xmath20 , and reduce @xmath19 .
the temperature does not offer a significant improvement unless the plates could be cooled far below liquid nitrogen temperature , and this does not seem practical in a molecular beam apparatus . larger plate spacing is also not a very useful option for the ybf as it would be challenging to maintain the required high electric field .
with a view to increasing @xmath20 , we have built field plates from commercial ito coated glass ( pilkington k glass ) , which has good vacuum properties and is transparent to visible light .
we find that the coating is prone to discharge , particularly at the edges .
this problem was also reported by @xcite , who tried to mitigate it by custom coating ito over rounded edges .
also , the surface is very easily damaged by discharges .
it has also been proposed that one might use soda glass , heated to a temperature of 150@xmath44c to achieve enough conductivity @xcite , but the addition of plate heaters is a complication we prefer to avoid in the ybf experiment .
our approach has therefore been to reduce @xmath19 , which we have done by making the electrodes from gold - coated polyimide ( kapton ) tape @xcite , on which the gold layer is only @xmath43 nm thick .
this is roughly @xmath45 of the previous plate thickness , while the @xmath46 m resistivity of gold is only slightly less than that of aluminium .
equation ( [ eq : bnzf0 ] ) suggest that this will reduce the magnetic field noise by approximately 100 , and numerical integration of eq.([eq : bnz ] ) shows this to be the case , as plotted by the red dashed line in fig.[fig : bnzplot ] . the corresponding phase uncertainty , given by eq.([eq : mungerd ] ) ,
is @xmath47rad for a 1 ms measurement in the ybf experiment , which converts to an uncertainty in the electron edm of @xmath48 e.cm over a data run of @xmath34 measurements .
this is sufficiently small that the new ybf experiment will not be limited by magnetic noise from the new plates .
we note that the magnetic noise from the glass substrates is negligible @xcite .
photograph and cross section of the film - on - glass electric field plates . ] these new , thin , field plates are built on slabs of float glass , with dimensions @xmath49 mm , chamfered on the edges at 45@xmath50 by a few mm , and rounded by hand using silicon carbide abrasive paper .
these are cleaned using detergent and water , then rinsed in distilled water followed by isopropanol .
a cold - roll laminator applies a continuous length of the self - adhesive , gold - coated polyimide tape to the front face of each plate , as shown in fig .
[ fig : plates](a ) .
excess tape is then folded over onto all four sides of the glass , as illustrated in cross - section in fig .
[ fig : plates](b ) .
when viewed in daylight the surface shows a ripple with a period of a few millimetres , but on measuring the height variations on a flat bed we found less than @xmath51 m peak - to - peak over a @xmath52 mm - square region and no perceptible ripple . on the top side , viewed in fig.[fig : plates](a )
, the tape is folded over to the back , where it is cut to form a small tab of @xmath53 mm , with its long axis parallel to the top edge of the substrate .
an aluminium block , shown in fig.[fig : plates](a ) , is glued over this tab using conducting epoxy , and is tapped to accept the threaded 2-mm end of a vacuum - compatible high - voltage cable .
the exposed edges of the gold - coated tape discharge at high voltage , but this is suppressed when we cover them with uncoated self - adhesive kapton tape . the four short corners are covered with a small quantity of _ torr seal _ vacuum epoxy to avoid any awkward edges . inside the vacuum chamber ,
the plates are mounted vertically along the beam line .
each plate is held by two horizontal alumina rods , glued with structural epoxy @xcite to the back face at the top and the bottom .
these plug into two aluminium support plates , where precisely machined holes set the plate separation .
the length of the leakage path from the gold conductor to the grounded aluminium support structure is 21 mm . once assembled and placed in the vacuum chamber ,
the entire apparatus is baked at 80@xmath44c for three days to accelerate the outgassing .
after cooling to room temperature , the pressure in the vacuum chamber falls to @xmath54 mbar , which is also the base pressure of the chamber without the plates .
we conclude that the outgassing rate from the tape with its acrylic adhesive is below @xmath0 mbar.l / s .
next , the plates are conditioned to support a large potential difference by applying a voltage and switching its polarity with about a one minute period .
we monitor the leakage current using na - sensitive leakage monitors @xcite and raise the voltage in 100v steps until current spikes of @xmath55 na start to appear . then the voltage is held constant ( while continuing to switch the polarity ) until the spikes disappear .
this takes between a few minutes and a few hours , after which the voltage can be further increased .
if the leakage current becomes continuous at greater than 10na the voltage is reduced for a time .
if large leakage currents ( @xmath56a ) are observed the voltage is reduced to zero and conditioning starts afresh . unlike the ito coating , the gold suffered no perceptible damage from these discharges , perhaps because the gold has much higher conductivity , both electrical and thermal . with this procedure ,
we reach a voltage of @xmath57kv after several days of conditioning .
the leakage current from the plates and high voltage cabling is @xmath58 na , which is dominated by leakage in the external 10m - long coaxial connecting cables and high voltage switching equipment .
the leakage current from the plates and high voltage connections under vacuum is less than 200 pa .
fractional variation of the electric field as a function of distance along the beam line , as measured by the stark shift of the hyperfine structure in ybf .
blue circles : aluminium plates .
red triangles : tape - on - glass plates . ]
it is important in the edm experiment that the molecules experience a homogeneous electric field between the plates @xcite .
we have mapped the field between the new plates at 4kv / cm , using the molecules themselves to sense the field strength through the stark effect of the hyperfine structure @xcite .
the triangular data points ( red online ) in fig .
[ fig : separation ] show the fractional variation of the field as a function of position along the length of the plates .
this is within the range @xmath59 , which provides a sufficiently uniform stark shift for the edm experiment . for comparison
, we also show the fractional field variation measured between the old plates , which is very similar in magnitude .
rotation of the electric field direction can also be undesirable when it induces a berry phase @xcite , but the variation we measure here occurs over a long distances and that implies a small rotation angle that is insufficient to produce a significant systematic error in the edm experiment .
it would also have been been possible to apply the gold layer by vacuum deposition , but the tape is considerably cheaper and allows field plates of any length to be produced .
it is also straightforward to replace the tape in - house if it were to become damaged by electrical discharge or surface contamination .
we acknowledge useful discussions with m. tarbutt . the research leading to these results has received funding from epsrc , stfc , the royal society and from the european research council under the european union s seventh framework programme ( fp7/2007 - 2013 ) / erc grant agreement 320789 .
n. solmeyer , _ progress towards an electron electric dipole moment measurement with laser - cooled atoms _ , phd thesis , penn .
state university ( 2013 ) .
section 6.4.4 .
available at https://etda.libraries.psu.edu/ _ sheldahl _ part number 169950 - 001 , polyimide thickness 25 @xmath10 m , gold thickness 90 nm , tape width 6 .
glue is acrylic 966 .
the kapton tape is 3 m polyimide film electrical tape 92 , 1 mil ( 25@xmath10 m ) kapton with silicone adhesive . | we describe a parallel pair of high voltage electric field plates designed and constructed to minimise magnetic johnson noise .
they are formed by laminating glass substrates with commercially available polyimide ( kapton ) tape , covered with a thin gold film . tested in vacuum , the outgassing rate is less than @xmath0 mbar.l / s .
the plates have been operated at electric fields up to 8.3 kv / cm , when the leakage current is at most a few hundred pa .
the design is discussed in the context of a molecular spin precession experiment to measure the permanent electric dipole moment of the electron . |
the evolution from starless to star - forming cores and further on to ( extra)solar systems is accompanied by a rich chemistry , which will affect planet and planetesimal compositions .
much of this chemical evolution takes place on interstellar grain surfaces , in icy mantles .
ices are generally observed in the cold and dense interstellar medium that are the nurseries of stars and are also abundant in our own solar system in comets and other outer solar system bodies .
such icy bodies are proposed to have delivered volatiles to the young earth .
understanding their link to the chemistry observed during star formation is key to constrain the amount of organic material delivered to earth as well as to exo - planets .
the first ices were detected in the interstellar medium almost 40 years ago by @xcite .
h@xmath0o and co were established early on as common ice constituents with abundances reaching 10@xmath5 @xmath6 .
this makes ices the most common molecules after h@xmath0 in many star forming regions .
co@xmath0 is the third major ice constituent . because of abundant atmospheric co@xmath0
, it was only detected in the ism after the launch of iras by @xcite .
following iras , the _ infrared space observatory _ ( iso ) and the _ spitzer space telescope _ have , together with ground - based efforts , continued the ice exploration and interstellar ices .
in addition to the major ice constituents , h@xmath0o , co and co@xmath0 are observed to often contain smaller amounts of ch@xmath7oh , ch@xmath8 , nh@xmath7 and ocn@xmath9/xcn ( see @xcite for a review ) .
the simple ices observed in the ism may become the source of complex molecules as described by @xcite and @xcite and determining ice abundances and production channels in star forming regions is of great interest for studies of prebiotic chemistry .
ices can form from direct freeze - out of molecules from the gas phase , through atomic addition on grain and ice surfaces , and through energetic processing of already existing ices . at the low temperatures ( @xmath220 k ) prevalent in cloud cores , atoms and molecules will stick to the grain upon impact . in this
stage grains are well protected from external uv radiation and atomic addition reactions probably dominate .
@xcite worked out in detail how the atoms and molecules can be hydrogenated and oxygenated to form all commonly observed ices , except for co , which condenses directly from the gas - phase . in this scheme ,
h@xmath0o forms from o+h , ch@xmath7oh from co+h etc .
ch@xmath7oh can also form from uv processing of h@xmath0o : ch@xmath8 ice mixtures , however , and there has been some debate on which formation pathway dominates . @xcite and others have also suggested that at least some co@xmath0 and xcn form from thermal or uv processing , mainly based on high - mass protostellar studies with _ iso_. _ spitzer _ and complimentary ground - based surveys have more recently allowed us to observe ices toward large numbers of low - mass protostars and cloud cores , which have provided new observational constraints on how ice formation and evolution depend on the local environment . the aim of this contribution is to review the constraints supplied by ice observations toward both _
iso _ and _ spitzer _ targets , on the identifications of ice bands and of the main formation paths of the identified carriers .
the figures are taken from berg et al .
( 2011 , subm . to apj ) unless otherwise noted .
l@xmath10 ) , hh46 ( 12 l@xmath10 ) and l1014 irs ( 0.09 l@xmath10 ) from @xcite and @xcite .
the 3 @xmath11 m portions of the spectra have been binned to increase the s / n.,width=384 ] ices are generally observed in absorption at mid - ir wavelengths and can therefore only be observed in lines of sight with infrared background sources .
protostars are excellent ir sources and ices in protostellar envelopes are the most extensively studied ( fig .
early ice observations were exclusively ground based and thus limited to observe ice bands that coincide with atmospheric windows .
iras and especially _ iso _ opened up the full infrared spectral region , resulting in the first complete ice inventories toward bright protostars and a few cloud lines of sight .
the _ iso _
ice results are described in @xcite and these still constitute the largest high - mass protostellar sample . compared to previous space missions , _ spitzer _ s high sensitivity enabled observations of ices toward low - mass protostars and also toward previously inaccessible background stars .
the largest sample was obtained within the @xmath12 program , defined in @xcite , which included irs spectra of 50 low - mass protostars .
the different ice bands were analyzed in a series of papers by @xcite , @xcite , @xcite and @xcite .
additional _ spitzer _
observations exist on ices toward background sources , looking through molecular clouds at a range of extinctions ( reported by @xcite , @xcite , @xcite , @xcite and @xcite ) and toward protostars in the high - uv environment of ic 1396a by @xcite . because the _ spitzer _ spectrometer had a lower cutoff at @xmath15 @xmath11 m , complete ice inventories
can only be obtained by adding complementary ground - based spectroscopy to cover the strong 3 @xmath11 m h@xmath0o , the 4.65 @xmath11 m co , the 4.60 - 4.62 @xmath11 m xcn , and the 3.53 @xmath11 m ch@xmath7oh features and this was done using vlt and keck .
results are summarized in berg et al .
( 2011 , subm .
to apj ) .
the combined _ spitzer _ , iso , vlt and keck ice sample spans the full range of observed luminosities from the low - luminosity object l1014 irs ( @xmath20.1l@xmath4 ) to the @xmath13 l@xmath4 massive protostar w33a ( fig .
[ fig1 ] ) .
the low - mass young stellar objects ( ysos ) are located in perseus , taurus , serpens , ophiuchus and corona australis , as well as a number of nearby isolated dense cores , and thus cover star - forming regions with different histories and physical environments .
the sample also represents a wide range of cloud , cloud core and yso evolutionary stages , from envelope - dominated class 0 sources to disks .
figure [ fig1 ] shows ice spectra toward protostars with the identifications of different bands to h@xmath0o , co , co@xmath0 , ch@xmath7oh , nh@xmath7 , ch@xmath8 and ocn@xmath14 ice marked .
@xcite found that the observed xcn band at 4.62 @xmath11 m consists of two components .
based on laboratory spectroscopy from @xcite , at least one of the components is due to ocn@xmath14 .
ocn@xmath14 in a separate ice matrix is probably the carrier of the second xcn band component ( centered at 2175 @xmath15 ) , but other identifications have also been suggested by e.g. @xcite .
oh feature in laboratory mixtures at 15 k ( left panel ) and in spitzer @xmath12 spectra ( right panel ) . in the right panel
an increasing symbol size is indicative of increasing ch@xmath7oh content . in both panels ,
the dash - dot polygons delimitate the parameter space of fwhm and peak positions corresponding to h@xmath0o - rich mixtures .
[ fig2],title="fig:",width=240 ] oh feature in laboratory mixtures at 15 k ( left panel ) and in spitzer @xmath12 spectra ( right panel ) . in the right panel
an increasing symbol size is indicative of increasing ch@xmath7oh content . in both panels ,
the dash - dot polygons delimitate the parameter space of fwhm and peak positions corresponding to h@xmath0o - rich mixtures .
[ fig2],title="fig:",width=240 ] in general the frequency of a solid state molecular vibration depends on the bonding environment of the vibrating molecule .
this results in that the positions and shapes of ice spectroscopic features depend on the ice mixture(s ) .
[ fig2 ] from @xcite demonstrates how the ch@xmath7oh fwhm and band center depend on the ice mixture in laboratory experiments and how this can be used to constrain that protostellar ch@xmath7oh ices are not present in h@xmath0o - dominated ices .
band on top of irs 51 , svs 4 - 5 , and rno 91 spectra from @xcite , width=192 ] some ice bands , such as co and co@xmath0 , can not be fitted by a single laboratory ice mixture , but seem to trace molecules in two or more ice phases , often referred to as polar and apolar .
traditionally the observed spectra have been directly compared to a superposition of laboratory pure and mixed ice spectra to constrain ice components , see e.g. @xcite , @xcite and @xcite .
the constraints are often degenerate , however , since ice spectral features vary with ice composition , temperature and radiation processing . to address this , observed ice bands are now commonly decomposed phenomenologically and then the components are compared to laboratory spectra , see @xcite , @xcite , @xcite and @xcite .
because all observed spectra are decomposed into the same , small number of components , this method provides information on the sample as a whole , i.e. it directly shows which parts of the spectral profile are ubiquitous and which are environment dependent .
figure [ fig3 ] from @xcite demonstrates how the derived co@xmath0 ice components can be identified with pure co@xmath0 ice , co@xmath0 mixed with co ( co@xmath0:co ) , co@xmath0 mixed with h@xmath0o ice ( co@xmath0:h@xmath0o ) and a shoulder associated with co@xmath0 mixed with ch@xmath7oh .
@xcite showed that the co ice band can be similarly decomposed into three components corresponding to pure co ice , co : co@xmath0 and co : h@xmath0o or co : ch@xmath7oh .
m spectra of two background stars ( bottom panels ) , compared to two ysos .
the sum of the components is indicated with a thick red line .
figure from @xcite .
_ right panel : _ spectra toward w33a , ngc7538 irs9 and b1-b , following subtraction of a local spline continuum , plotted together with laboratory spectra of hcooh , ch@xmath7cho and ch@xmath7ch@xmath0oh ices .
the dashed lines mark the 7.25 and 7.40 @xmath11 m features usually assigned to hcooh and hcoo@xmath14 . , title="fig:",width=259 ] m spectra of two background stars ( bottom panels ) , compared to two ysos .
the sum of the components is indicated with a thick red line .
figure from @xcite .
_ right panel : _ spectra toward w33a , ngc7538 irs9 and b1-b , following subtraction of a local spline continuum , plotted together with laboratory spectra of hcooh , ch@xmath7cho and ch@xmath7ch@xmath0oh ices .
the dashed lines mark the 7.25 and 7.40 @xmath11 m features usually assigned to hcooh and hcoo@xmath14 .
, title="fig:",width=244 ] figure [ fig1 ] also shows tentative identifications to hcooh and nh@xmath16 in a complex band between 5 and 7 @xmath11 m . to constrain the carriers of this band @xcite
decomposed it into five different components ( c1c5 ) after subtraction of the contribution from h@xmath0o ice ( fig .
[ fig4 ] ) . from comparison with laboratory ice spectra ,
c1 has been identified with hcooh and h@xmath0co , c2 with hcoo@xmath14 and nh@xmath7 , c3 with nh@xmath16 and ch@xmath7oh , c4 with nh@xmath16 and c5 with warm h@xmath0o and anions .
the 7.25 @xmath11 m has been used to derive hcooh ice abundances by e.g. @xcite , @xcite and @xcite .
the right panel of figure [ fig4 ] shows that ch@xmath7ch@xmath0oh is an equally plausible carrier , however .
the presence of a 7.25 @xmath11 m feature is evidence for the formation of complex ices , but derivations of hcooh abundances from this band is not possible without higher resolution spectra that can distinguish between the two carriers .
following ice identification , ice column densities are determined by dividing the integrated optical depth of the ice feature by the band strength .
ice abundances are expressed with respect to h@xmath0o ice , because h@xmath0o ice is predicted to form early during star formation , is the most abundant ice , and has a high sublimation point .
determining ice abundances is important to predict the chemical evolution in different environments and to constrain the c , o and n budget .
in addition to these direct uses , the following sections demonstrate how ice abundance statistics can be used to put constraints on how ices form .
comparison of ice distributions between clouds and protostars constrain which ices require protostellar heating or uv processing to form .
ice abundance variations constrain which ices co - form with h@xmath0o and which ices form in a separate stage .
ice correlations provide further constraints on co - formation , and anti - correlations can be used to demonstrate competitive formation . in these investigations
it is key to quantify how significant differences and correlations really are .
finally , ice maps provide direct evidence of how ice formation depends on the position of a source in a cloud or a core .
.recommended ice abundances toward low- and high - mass protostars and clouds .
[ tab : median ] [ cols="<,^,^,^",options="header " , ] and ch@xmath7oh toward low - mass protostars ( red fill ) , our background star sample ( black line - fill ) and taurus background stars ( blue contours).,width=288 ] comparisons between protostars and background stars provide direct limits on which ices form before the protostar turns on and starts to heat and irradiate its surroundings .
@xcite showed that in taurus the co@xmath0 ice fraction in clouds is lower compared to low - mass protostellar envelopes , and proposed that at least some of the co@xmath0 observed toward protostars forms from energetic processing . similarly
, ch@xmath7oh ice has been proposed to be a product of protostellar uv ice - processing by e.g. @xcite .
figure [ fig5 ] shows that the co@xmath0 and ch@xmath7oh ice abundances toward protostars and background stars from @xcite are similar , and that the taurus co@xmath0 ice abundances from @xcite are poor templates for ice abundances toward average cloud cores .
@xcite show that there is also no statistically significant difference in the c1 - 4 components between the protostars and background stars , but the c5 component is absent in cloud sources.the only other feature that is present exclusively toward protostars is pure co@xmath0 ice , which is proposed to form from thermally induced ice segregation or distillation .
the protostar thus seems to have a very limited impact on the observable ice composition .
this entails that protostellar ice abundances mainly reflect the ice formation process in the previous cold cloud phase .
ice abundance variations between different sources depend on how sensitive the ice formation and destruction pathways are to the local environment
. since ice abundances are with respect to h@xmath0o , a small variation of a species indicates co - formation with h@xmath0o , while large ice abundance variations are indicative of different formation and/or destruction dependencies than h@xmath0o ice .
figure [ fig6 ] shows that the co@xmath0 , ch@xmath8 and nh@xmath7 abundance distributions are narrow , indicative of that these ices form together with h@xmath0o , while the co , ocn@xmath14 and ch@xmath7oh distributions are broad .
the ice abundances around high - mass protostars are generally lower compared to low - mass protostars , indicating that heat and uv mainly destroy simple ices .
the large ch@xmath7oh , co and ocn@xmath14 abundance variations must then originate from a separate prestellar ice formation stage compared to the stage associated with h@xmath0o formation .
this is consistent with observations that most prestellar and protostellar co@xmath0 is present in a h@xmath0o rich ice , while ch@xmath7oh and co ices are not .
figure [ fig6 ] shows the distributions of the individual co and co@xmath0 components .
the pure co and co@xmath0 ice components have the broadest distributions , consistent with their expected dependence on the protostellar envelope temperature for ice evaporation and segregation .
all other co and co@xmath0 component distributions , except for co@xmath0:h@xmath0o , are broad as well , indicative of a different history compared to the h@xmath0o - rich ice phase .
m map from @xcite with respect to the central core ( star ) .
_ right panel : _ ice abundances at different distances from the oph - f core with respect to h@xmath0o ice .
the 2175 @xmath15 abundances are scaled by 40 and the co@xmath0:co abundances by 5 for clarity .
[ fig7],title="fig:",width=312 ] m map from @xcite with respect to the central core ( star ) .
_ right panel : _ ice abundances at different distances from the oph - f core with respect to h@xmath0o ice .
the 2175 @xmath15 abundances are scaled by 40 and the co@xmath0:co abundances by 5 for clarity .
[ fig7],title="fig:",width=264 ] the spatial extent of ices can be mapped in clouds and cloud cores if enough background sources are available .
mapping of ice abundances in dense clouds by @xcite has revealed a similar formation threshold of h@xmath0o and co@xmath0 at a@xmath17=4 mag .
co is only observed at higher extinctions .
the h@xmath0o - rich ice thus forms first .
@xcite shows that toward the oph - f core , where protostars are used as background sources to probe the cloud core ices , co ice abundances increase dramatically toward the core ( fig .
[ fig7 ] ) .
@xcite interpreted this as catastrophic freeze - out of co in the pre - stellar stage once a certain density and temperature is reached .
figure [ fig7 ] shows that the order of magnitude increase in co ice with respect to h@xmath0o toward the core is accompanied by an increase in co@xmath0:co .
in contrast , the co@xmath0:h@xmath0o ice component is almost constant .
co : h@xmath0o and xcn band are the only other species that increase monotonically toward the densest part of oph - f core .
these bands thus appear directly related to co freeze - out , indicating that their broad abundance distributions in fig .
6 are due to their dependence on the prestellar co freeze - out level .
no trend with distance to the core center is seen or expected for ices that form with h@xmath0o ( e.g. ch@xmath8 and nh@xmath7 ) , or of species dependent on protostellar heating ( e.g. pure co@xmath0 ice ) or of components with potentially multiple carriers , such as the c1 - 5 bands .
ch@xmath7oh is only detected toward one of the sources and no trend can be extracted .
o ( n@xmath18 / % n@xmath19 ) and the h@xmath0o ice column density for low - mass protostars , with upper limits plotted with thin symbols .
the significance of the correlations was measured using spearman s rank correlation test , which makes no assumptions about the type of correlation , while the dotted line shows the best log - log fit to the data to guide the eye.[fig8],width=384 ] ice maps are extremely useful , but only possible in a few lines of sight with fortuitous placement of background sources .
ice correlation tests is another tool used to constrain the ice formation history , used by e.g. @xcite to investigate which ice components depend on ice heating to form .
figure [ fig8 ] shows an example of such a correlation study using the @xmath12 low - mass protostellar sample . among the investigated ices , the abundance w.r.t h@xmath0o of ch@xmath8 , co@xmath0:co , the co@xmath0 shoulder , co : h@xmath0o , co : co@xmath0 and c4 are anti - correlated with the h@xmath0o column density at 9799% confidence with the rank correlation test introduced by @xcite . the co - related anti -
correlations are consistent with the 4.5 results that h@xmath0o - rich and a co - rich ice phases form separately in clouds and cloud cores .
the anti - correlations are expected if there is competition between the formation of these two phases ; i.e. the more oxygen that is bound up in h@xmath0o ice , the less may be available to form co and thus ices that depend on co freeze - out . still in each individual core co ice abundance and h@xmath0o ice column
are expected to correlate because of increasing co freeze - out toward the densest part of the cloud core .
this accounts for some of the scatter in the plots .
the ch@xmath8h@xmath0o trend is strongest for low h@xmath0o column densities and may be due to more efficient ch@xmath8 formation during the earliest h@xmath0o formation stage when then c / co ratio is high .
building on the presented results above as well as additional experiments , models and observations , this section discusses what the current constraints on the ice evolution from clouds to protostars .
o dominated ice with a high co@xmath0 concentration ( co@xmath0:h@xmath0o ) . at later times co freezes out catastrophically , resulting in a second layer where co@xmath0 formation continues ( co@xmath0:co ) and ch@xmath7oh formation begins .
small amounts of ice surface chemistry products are maintained in gas - phase due to non - thermal desorption . in the protostellar stage desorption of co
starts at 20 k and h@xmath0o : co@xmath0 segregation at 30 k. [ fig11],width=384 ] from ice maps of cloud edges , the first ices to form are h@xmath0o dominated with large concentrations of co@xmath0 ( fig .
ch@xmath8 formation is the most efficient at the earliest stages of h@xmath0o ice formation , when the c / co ratio in the gas phase is still high ( 4.6 ) .
the narrow abundance distribution of nh@xmath7 is evidence for that it too forms with h@xmath0o .
this ice formation stage depend on hydrogenation of atoms , o+h , c+h and n+h ; co@xmath0 can form efficiently from co+oh at low temperatures , based on experiments by @xcite , @xcite and @xcite .
most co that is frozen out in this phase must be converted into co@xmath0 since co@xmath0:h@xmath0o is more abundant than co : h@xmath0o . from experiments by @xcite
, h preferentially reacts with atoms compared to co , explaining why no ch@xmath7oh forms in this ice formation phase . at some point
the co / o ratio becomes large enough that most frozen - out co is no longer converted into co@xmath0 and h@xmath0o is no longer the most efficiently formed ice .
we mark this as the breaking point between early h@xmath0o dominated and late co - driven pre - stellar ice formation .
the ices that form in this later stage are not expected to co - vary with h@xmath0o , but rather depend on the co - freeze - out efficiency .
the latter is both temperature and density dependent and may vary greatly dependent on cloud structure , cloud collapse time scales and local radiation environment .
this explains why many identified ice and ice components that are definitely present during the pre - stellar stage have abundances that vary by orders of magnitude with respect to h@xmath0o .
in addition , the anti - correlation between h@xmath0o ice and the abundance of several co - ice components suggest that the time available for this co - based chemistry decreases when a longer time is spent in the h@xmath0o - ice formation stage , where o is converted into h@xmath0o and co@xmath0:h@xmath0o ice .
co@xmath0:co , co : h@xmath0o and the xcn band all correlate with co freeze - out in the oph - f ice map .
co@xmath0 may still form from oh+co at this stage , but a high co / o ratio results in sparse h@xmath0o formation compared to co and thus in a co - rich co@xmath0 ice .
this is also the outcome from recent modeling by @xcite . ocn@xmath9
may form from co+nh , followed by proton transfer , and the xcn - co freeze - out correlation supports that ocn@xmath14 is responsible for the entire xcn band . @xcite
have used recent experiments to show that ch@xmath7oh and h@xmath0co can form from hydrogenation of co under ism conditions and ch@xmath7oh abundance should also depend on co freeze - out .
in addition to atomic processing , the ices are continuously exposed to cosmic rays and cosmic - ray induced uv photons @xcite have estimated the relative doses .
this may both affect the simple ice abundances and result in some cold formation of more complex molecules . though much of the observed ice evolution can be explained by processes during the prestellar stage ,
some ice features are formed or destroyed by the protostar .
based on the observations of pure co@xmath0 by @xcite and experiments by e.g. @xcite , evaporation of co from co@xmath0:co ice and segregation of h@xmath0o : co@xmath0 ices are efficient above 2030 k in protostellar envelopes .
pure co@xmath0 is also observed by @xcite around extremely low - luminosity protostars , however , and may there be a remnant of past outbursts that heated up the envelope beyond its low - accretion temperature , since ice segregation is an irreversible process . @xcite and
@xcite have invoked episodic accretion to explain the luminosity problem of low - mass ysos , i.e. the observation that most ysos are significantly less luminous than predicted by protostellar evolutionary models with steady accretion . the observed lower co and ch@xmath8 ice abundances toward high - mass protostars compared to low - mass ysos can also be explained by protostellar ice heating since these are very volatile ices . from experiments by
e.g. @xcite , @xcite and @xcite , some co and ch@xmath8 can be trapped in h@xmath0o ices , however , which is important for predicting ice abundances in moderately heated regions , such as the comet forming zones in protoplanetary disks . from experiments by e.g. @xcite
, moderate ice heating to 2050 k also enables the movement of radicals within the ice .
radicals are produced in ices by photodissociation of the condensed molecules .
models by @xcite predict that especially ch@xmath7oh photodissociation followed by radical diffusion result in complex ice formation and protostellar envelopes are thus expected to be efficient factories of complex organics .
uv radiation of co : ch@xmath7oh ice mixtures produce a more hco - rich complex chemistry , which may dominate at lower temperatures when co is still frozen out .
the ice feature at 7.25 @xmath11 m identified with to hcooh or ch@xmath7ch@xmath0oh provides direct evidence for this complex ice chemistry .
most complex ice features can not be directly observed , however , because of band confusion .
instead most constraints on this ice evolutionary phase comes from millimeter observations of desorbed ice chemistry products .
complex organic molecules such as ch@xmath7ch@xmath0oh and hcooch@xmath7 are observed in the inner regions around high - mass protostars , in so called hot cores , at abundances that can not be reproduced by gas phase chemistry .
furthermore , @xcite found that they correlate with the ice chemistry product ch@xmath7oh .
these observational pieces of evidence , together with the existence of a plausible ice production pathway of complex molecules , strongly support a scenario in which complex ices are produced in the outer envelope and then desorb into the gas - phase as icy grains flow in towards the protostar .
iras 16293 - 2422 , observed by e.g. @xcite and @xcite is a low - mass equivalent to the high - mass hot cores , and there are at least five other low - mass protostars where complex molecules such as hcooch@xmath7 and ch@xmath7och@xmath7 are detected by @xcite , @xcite and @xcite .
@xcite also detected complex molecules in a low - mass outflow and @xcite recently detected hcooch@xmath7 in a cold and quiescent part of a low - mass star forming region that probably has been illuminated by uv through a protostellar outflow cavity . combining these observational results ,
berg et al . ( 2011b , subm .
to apj ) find that hcooch@xmath7 generally dominates the complex chemistry toward outflows and low - mass protostellar envelopes , consistent with experiments on co : ch@xmath7oh ice photochemistry below 25 k. in contrast , observations of high - mass and resolved low - mass hot cores contain more ch@xmath7och@xmath7 or c@xmath0h@xmath20oh than hcooch@xmath7 .
this is evidence for a sequential formation of complex molecules , starting with hco - rich molecules as long as co ice is abundant , followed by ch@xmath21-rich molecules at higher ice temperatures .
gas phase observations can thus provide information on the ice chemistry , but interpretations of these more indirect constraints rely on a combination of experimental and model results .
thanks to a combination of ground - based and space telescope ice observations and years of laboratory ice spectroscopy and ice chemistry experiments we have good constraints on the typical ice abundances and ice compositions in a range of interstellar and circumstellar environments . a key result coming out of these observations
is that most ices up to ch@xmath7oh in complexity have already formed by the time the protostar turns on .
ice abundance variations must therefore be mainly due to variations in the prestellar ice formation processes .
variations in co freeze - out and the subsequent co - based chemistry is a likely main contributor to the observed differences in ice abundances in different lines of sight , but other processes may be important as well
. thermal and uv processing of ices by the protostar is important for the continued evolution of the ice , leading up to the formation of hot core molecules and the complex organics found in comets . to further constrain the ice evolution from cloud cores to comets
will require a combination of high - resolution ir studies to detect complex molecules directly in the ice , millimeter observations toward protostars and disks to image desorbed ice chemistry products , and more detailed experiments and models of both the early ice formation stages and of the reactions that convert simple ices into more complex ones . | icy grain mantles are commonly observed through infrared spectroscopy toward dense clouds , cloud cores , protostellar envelopes and protoplanetary disks .
up to 80% of the available oxygen , carbon and nitrogen are found in such ices ; the most common ice constituents
h@xmath0o , co@xmath0 and co are second in abundance only to h@xmath0 in many star forming regions .
in addition to being a molecular reservoir , ice chemistry is responsible for much of the chemical evolution from h@xmath0o to complex , prebiotic molecules . combining the exisiting iso , spitzer , vlt and keck ice data results in a large sample of ice sources ( @xmath180 ) that span all stages of star formation and a large range of protostellar luminosities ( @xmath20.110@xmath3 l@xmath4 ) .
here we summarize the different techniques that have been applied to mine this ice data set on information on typical ice compositions in different environments and what this implies about how ices form and evolve during star and planet formation .
the focus is on how to maximize the use of empirical constraints from ice observations , followed by the application of information from experiments and models .
this strategy is used to identify ice bands and to constrain which ices form early during cloud formation , which form later in the prestellar core and which require protostellar heat and/or uv radiation to form .
the utility of statistical tests , survival analysis and ice maps is highlighted ; the latter directly reveals that the prestellar ice formation takes place in two phases , associated with h@xmath0o and co ice formation , respectively , and that most protostellar ice variation can be explained by differences in the prestellar co ice formation stage .
finally , special attention is paid to the difficulty of observing complex ices directly and how gas observations , experiments and models help in constraining this ice chemistry stage . |
einstein gravity is the most successful theory to describe gravitation . however ,
if we include quantum effects , einstein gravity can not be a fundamental theory , and hence a quantum theory of gravity is required .
this quantum gravity should be approximated by the einstein gravity in the low energy limit , but before approaching the einstein gravity , there may be some effects that modify the einstein gravity .
in fact , a theory of gravity does not definitively have to be einstein gravity .
there can be many variations and we choose one theory via observations .
we then need to find possible modified gravity theories and check their possible implications .
if quantum gravity is approximated by a modified gravity theory , and if the theory has experimental implications , it will be profoundly important to understand the nature of quantum gravity .
one of the most canonical modified gravity theory candidates is the brans - dicke theory @xcite : @xmath4 where @xmath5 is the ricci scalar , @xmath6 is the brans - dicke field , and @xmath7 is the brans - dicke coupling constant . historically , the brans - dicke theory was suggested to implement mach s principle .
the roll of the gravitation constant @xmath8 is replaced by the brans - dicke field @xmath9 .
since the brans - dicke field can be dynamic , the strength of gravity can be relatively different for each points .
it makes that the equivalence principle has more freedom than the original einstein gravity , since the equivalence principle holds for each points independently .
then , intuitively , we can recover the einstein gravity in the large @xmath10 limit , since the field dynamics of the brans - dicke field will be highly suppressed in the limit .
moreover , the brans - dicke gravity has other theoretical motivations .
notably , it appears that it is the simplest form of modified gravity using a scalar field .
therefore , some important intuitions of the brans - dicke theory can be applied to other possible modified gravity theories .
for example , we may apply similar intuitions to dilaton gravity , non - minimally coupled theory , etc .
the brans - dicke theory can be motivated by string theory inspired models .
for example , the brans - dicke theory is the weak coupling limit of dilaton gravity ( @xmath11 ) , where the dilaton field is a direct consequence of string theory @xcite . also , the brans - dicke theory is a weak field limit of the randall - sundrum model @xcite , where , on the positive tension brane
, @xmath7 is sufficiently large ; on the negative tension brane , @xmath12 @xcite@xcite .
of course , the observed limitation of the coupling is @xmath13 @xcite ; however , it is still meaningful to study other values of @xmath7 since they can be realized in the context of string theory , and hence they can show properties which are allowed by string theory
. for further discussions , see appendix a. if the brans - dicke field is added to einstein gravity , it can work as a source of an exotic matter . in the context of cosmology , there have been some discussions that the brans - dicke field can be a candidate for dark matter or dark energy @xcite . in particular , it can be useful to study exotic matters that violates the null energy condition .
the theory can allow some strange geometry , e.g. , wormholes @xcite .
therefore , if we know how to deal with the brans - dicke field so as to obtain a sufficient amount of exotic matters , then we can reasonably make an assumption of exotic matter , since the exotic matter may thus be naturally obtained from a well - known and not artificial field
. although the study of the brans - dicke field is important , solutions to the brans - dicke theory are more difficult than those involving einstein gravity . of course
, we can find some analytic solutions .
however , to observe the real dynamics when a black hole is formed and evolved , more detailed calculations are needed .
the collapsing matter will affect the background , and the background will affect the brans - dicke field .
the affected brans - dicke field will then give a back - reaction to the background geometry .
these processes are not easy to understand in an analytic way and it is necessary to use numerical calculations for their elucidation .
previous papers have studied brans - dicke black holes using adm formalism @xcite . however , in this paper , we use double - null formalism @xcite@xcite@xcite for the numerical calculations @xcite@xcite , thus making it easier to study the causal structures and also to plot all possible fields including the energy - momentum tensors .
the purpose of this paper is to study responses of the brans - dicke field during the gravitational collapse of a matter .
we prepare a flat background with a pulse of a normal scalar matter field and we then observe the response of the brans - dicke field .
we thereupon observe that dynamical responses of the brans - dicke field are highly non - trivial .
this paper is organized as follows : in section [ sec : mod ] , we discuss the details of our numerical simulations ; in section [ sec : res ] , we discuss the results of our simulations and report some interesting observations ; in section [ sec : inter ] , we document observations in terms of two issues ( late time behaviors and dynamical behaviors ) and discuss their physical interpretations ; and in section [ sec : con ] , we summarize the possible causal structures and discuss the physical meanings and implications .
the lagrangian of the brans - dicke theory with a scalar field becomes @xcite @xmath14 where @xmath5 is the ricci scalar , @xmath6 is the brans - dicke field , and @xmath15 is a minimally coupled scalar field with potential @xmath16 . here , @xmath7 is the brans - dicke coupling constant which is a free parameter of the theory .
the einstein equation becomes as follows : @xmath17 where the brans - dicke part of the energy - momentum tensors are @xmath18 and the matter part of the energy - momentum tensors are @xmath19 the field equations are as follows : @xmath20 where @xmath21 we use the double - null coordinates @xmath22 assuming spherical symmetry , where @xmath23 is the retarded time , @xmath24 is the advanced time , and @xmath25 and @xmath26 are angular coordinates .
we follow the notation of @xcite@xcite@xcite@xcite : the metric function @xmath27 , the radial function @xmath28 , the brans - dicke field @xmath6 , and a scalar field @xmath29 , and define @xmath30 the einstein tensors are then given as follows : @xmath31 also , we can obtain the energy - momentum tensors for the brans - dicke field part and the scalar field part : @xmath32 @xmath33 to implement double - null formalism into the numerical scheme , it is convenient to represent all equations as first order differential equations . note that @xmath34 the einstein equations for @xmath35 , @xmath36 , and the field equation for @xmath6 are then coupled : @xmath37 where @xmath38 after solving these coupled equations , we can write all equations : @xmath39 including the scalar field equation @xmath40 now , equations of @xmath35 , @xmath36 , @xmath41 , and @xmath42 parts can be represented by first order differential equations .
we can then implement the same integration scheme used in previous papers @xcite@xcite to solve the brans - dicke theory .
we used the second order runge - kutta method @xcite .
tests of the convergence are provided in appendix b. we need initial conditions for all functions ( @xmath43 ) on the initial @xmath44 and @xmath45 surfaces , where we set @xmath46 .
we have gauge freedom to choose the initial @xmath28 function .
although all constant @xmath23 and @xmath24 lines are null , there remains freedom to choose the distances between these null lines .
here , we choose @xmath47 , @xmath48 , and @xmath49 , where @xmath50 and @xmath51 such that the radial function for an in - going observer decreases and that for an out - going observer increases .
first , we assume that the gravitation constant @xmath52 is asymptotically @xmath53
. then , @xmath54 and @xmath55 .
we use a shell - shaped scalar field , and hence its interior is not affected by the shell .
thus , we can simply choose @xmath56 and @xmath57 .
also , @xmath58 holds .
then , since the asymptotic mass function @xcite @xmath59 should vanish at @xmath60 , it is convenient to choose @xmath61 and @xmath62 .
we need more information to determine @xmath63 , and @xmath64 on the @xmath65 surface .
we obtain @xmath66 from equation ( [ eq : e3 ] ) , @xmath67 from equation ( [ eq : e4 ] ) , @xmath68 from equation ( [ eq : s ] ) , and @xmath64 from equation ( [ eq : s ] ) , respectively .
we can choose an arbitrary function for @xmath69 to induce a collapsing pulse . in this paper , we use @xmath70 for @xmath71 and otherwise @xmath72 , where @xmath73 is the width of the pulse and @xmath74 is the amplitude . then we obtain @xmath75 for @xmath71 and otherwise @xmath76 .
this implements one pulse of energy ( @xmath77 ) along the out - going null direction by a differentiable function @xmath78 .
also , from equation ( [ eq : e2 ] ) , we can use @xmath79 on the @xmath80 surface and we obtain @xmath81 . by integrating @xmath66 along @xmath24
, we have @xmath82 .
we need more information for @xmath83 and @xmath84 on the @xmath80 surface .
we obtain @xmath85 from equation ( [ eq : e3 ] ) , @xmath86 from equation ( [ eq : e4 ] ) , @xmath87 from equation ( [ eq : s ] ) , and @xmath84 from equation ( [ eq : s ] ) , respectively .
this finishes the assignments of initial conditions .
we choose @xmath88 , leaving the three free parameters @xmath89 , where @xmath7 is the brans - dicke coupling parameter , @xmath74 is the amplitude of a pulse of the scalar field , and @xmath73 is the width of the pulse . here
, we assume that there is no potential term in the matter field side .
contour diagrams of the radial function @xmath28 .
@xmath90 , @xmath91 , and @xmath92 .
spacing of each contour is @xmath53 . here , we plotted singularities , horizons ( @xmath93 and @xmath94 ) , and cauchy horizons . ]
we report the simulation results in figure [ fig : varying_omega ] .
we fixed @xmath90 and @xmath91 , and varied @xmath95 . note that if @xmath96 , because of equation ( [ eq : phi ] ) , the equation becomes singular .
figure [ fig : varying_omega ] contains contour diagrams of the radial function @xmath28 for each parameters .
we plotted singularities , trapping ( apparent ) horizons , and cauchy horizons . *
if the radial function becomes @xmath97 , since the @xmath98 curve is space - like and all equations become singular , it is reasonable to interpret this region as a singularity . * to define a black hole using local geometry , we may use apparent horizons @xcite@xcite or trapping horizons @xcite for an out - going observer , i.e. , @xmath93 .
if there is inflation at some point in space - time , the inflating region can be defined by an in - going observer , i.e. , @xmath94 , so that the in - going observer sees an increase of the area function .
we plotted two horizons : @xmath93 and @xmath94 . *
if an apparent horizon contracts to a singularity , we can find a null cutoff line where we can not calculate , because we do not have sufficient information to determine this region via the singularity ; hence , we can identify the region as a cauchy horizon @xcite@xcite .
the integration domain of the double - null simulation .
region ( a ) is for the late time behavior ; ( b ) is for the dynamical behavior . ] in this subsection , we list some interesting observations for figure [ fig : varying_omega ] . in the next section ,
we interpret the observations . 1 .
eventually a black hole is formed , i.e. , a space - like singularity and an apparent horizon are observed .
if @xmath10 is sufficiently large , the causal structures become similar to each other and the apparent horizons are always space - like .
see @xmath99 and @xmath100 cases .
3 . as @xmath7 decreases , if @xmath101 , the apparent horizon approaches the event horizon in the time - like direction . see @xmath102 .
4 . if @xmath103 and @xmath7 is sufficiently small , the apparent horizon oscillates and eventually approaches the event horizon in the space - like direction .
see @xmath104 . 5 .
if @xmath12 , the causal structure can contain an instantaneously inflating region , i.e. , there can exist an @xmath94 horizon .
as @xmath7 approaches @xmath0 , the @xmath94 horizon appears from upper @xmath23 to lower @xmath23 .
compare @xmath11 and @xmath105 .
if @xmath106 , the apparent horizon can shrink to a singularity and form a cauchy horizon . however , eventually a black hole will be formed .
see @xmath105 .
observations @xmath107 and @xmath108 pertain to asymptotic behaviors after the collapse of matter .
we call them _ late time behaviors_. observations @xmath109 and @xmath110 pertain to fully dynamical behaviors during the gravitational collapse .
we call them _ dynamical behaviors _ ( figure [ fig : domain ] ) .
we discuss and interpret the late time behaviors and dynamical behaviors in the following section .
from the work of previous authors using brans - dicke theory , we know that a gravitational collapse should lead to the emergence of an einstein black hole asymptotically @xcite . as the black holes emerge to einstein black holes , some interesting observations can be made . to understand these observations , in this section
, we discuss the large @xmath10 case and the @xmath111 case independently .
plot of the brans - dicke field @xmath6 .
@xmath90 , @xmath91 , and @xmath92 .
if the color changes , it shows that the brans - dicke field has dynamical behaviors .
if @xmath112 , then @xmath6 decreases during the matter collapse ; if @xmath113 , then @xmath6 increases during the matter collapse . ]
if @xmath10 is sufficiently large , the field equation of the brans - dicke field ( equation ( [ eq : phi ] ) ) becomes a free scalar field equation .
then , if @xmath6 was asymptotically @xmath53 and initially had no dynamics , it will not be affected by gravitation , since the field equation is a free scalar field equation .
( see the cases of @xmath99 and @xmath114 in figure [ fig : sb ] . ) therefore , the dynamics during the collapse of matter will be similar to that of einstein gravity and we can ignore the effect of the brans - dicke field .
this explains why the black holes in the large @xmath10 limit resemble einstein black holes .
note that , in these cases , space - like apparent horizons are observed .
if there is no supply of matter , the horizon should be null .
however , during the matter collapse , there is some scattered energy along the out - going direction , and part of this energy will come to the black hole later .
this explains why the horizons are still space - like and not null after @xmath91 .
plot of the scalar field @xmath115 .
@xmath90 , @xmath91 , and @xmath116 and @xmath117 .
these figures show that , during collapse of matter , @xmath115 decreases for both the in - going and out - going directions .
therefore , @xmath118 holds during matter supply . for other parameters ,
the behaviors are similar during the matter collapse . ]
if @xmath7 is near @xmath0 , we will see dynamical effects of the brans - dicke field .
figure [ fig : sb ] shows that if @xmath7 is greater than @xmath0 , it decreases during the matter collapse ; if @xmath7 is less than @xmath0 , it increases during the matter collapse .
we discuss and interpret these phenomena below .
we have the equation for the brans - dicke field : @xmath119 during the matter collapse , the @xmath120 term is dominant . also , in our initial condition , @xmath120 is always positive during the matter collapse ( figure [ fig : sm ] ) .
then , the total sign of @xmath41 is determined by the sign of @xmath121 .
therefore , if @xmath112 , the total sign of @xmath41 becomes negative , and hence the brans - dicke field tends to decrease .
if @xmath113 , by the same reasoning , the brans - dicke field tends to increase .
after the matter supply ends , the brans - dicke field will be relaxed to the asymptotic value , since the black hole should approach an einstein black hole .
hence , if @xmath112 , the decreased brans - dicke field tends to increase for an out - going observer around the horizon .
note that the area function in the einstein frame is @xmath122 @xcite , where @xmath123 is the area of a black hole in the jordan frame .
we know that , according to the area theorem , the area of the einstein frame ( @xmath122 ) will always increase and be space - like .
then , @xmath124 if there is no collapse of additional matter , @xmath125 and @xmath126
. therefore , if @xmath6 increases along the horizon , the area should shrink and be time - like .
if @xmath113 , by the same reasoning , we can explain that the horizon contracts along a space - like direction .
now we turn to a discussion of fully dynamical behaviors during the matter collapse .
we have observed interesting behaviors of @xmath93 or @xmath94 horizons .
these behaviors should be consistent with energy - momentum tensors .
therefore , in this section , we look at details of the energy - momentum tensors and energy conditions that cause the dynamical behaviors . we discuss two different cases , @xmath12 and @xmath106 . to understand the dynamical behaviors , we plot the energy - momentum tensor components ( figure [ fig : lesst ] ) . for any observer
whose four velocity is @xmath127 , the observer feels the energy density by @xmath128 therefore , if an observer moves along an in - going null direction , the observer feels the energy density by @xmath2 ; if an observer moves along an out - going null direction , the observer feels the energy density by @xmath3 .
note that the null energy condition is violated if @xmath2 or @xmath3 has a negative region .
around horizons ( @xmath94 or @xmath93 ) , we can simplify equations ( [ eq : e1 ] ) and ( [ eq : e2 ] ) by @xmath129 we know that asymptotically @xmath130 and @xmath131 ; i.e. , initially , an in - going observer sees a decrease of area and an out - going observer sees an increase of area . using these observations , we remark on two cases where the null energy condition should be violated .
* we can define the outer part of an @xmath94 horizon if the sign of @xmath132 changes from @xmath133 to @xmath134 along an in - going null direction ; the inner part of an @xmath94 changes its sign from @xmath134 to @xmath133 along an in - going null direction .
then , around the outer @xmath94 horizon , @xmath135 should be positive , and hence @xmath136 . *
if @xmath137 around the ( outer ) @xmath93 horizon , the horizon is space - like ; if @xmath138 around the @xmath93 horizon , the horizon is time - like . therefore , to see a time - like @xmath93 horizon , @xmath139 is required
note that , @xmath140 component is related to the vacuum energy or cosmological constant term .
therefore , if @xmath141 , then it may cause an increase of the area ; if @xmath142 , then it may cause a decrease of the area .
we can write equation ( [ eq : e4 ] ) around the @xmath94 horizon as @xmath143 if there is inflation , an out - going null observer sees an increase of area .
this requires that @xmath144 , and hence @xmath140 should be sufficiently large : @xmath145 if the inflation tends to end , the opposite situation should occur and @xmath140 should be sufficiently small so that @xmath146 .
therefore , we need to check the sign of @xmath140 components if there is an @xmath94 horizon . the energy - momentum tensors for ( @xmath147 , @xmath90 , @xmath91 ) .
] the energy - momentum tensors of the matter sectors ( @xmath148 , @xmath149 ) and the brans - dicke sectors ( @xmath150 , @xmath151 ) for ( @xmath147 , @xmath90 , @xmath91 ) . ]
the energy - momentum tensors for ( @xmath105 , @xmath90 , @xmath91 ) . ]
the energy - momentum tensors of the matter sectors ( @xmath148 , @xmath149 ) and the brans - dicke sectors ( @xmath150 , @xmath151 ) for ( @xmath105 , @xmath90 , @xmath91 ) . ] in figure [ fig : lesst ] , we can see the energy - momentum tensor components of the @xmath12 case .
the outer @xmath94 horizon was possible since the @xmath136 condition holds .
also , @xmath93 could bend in the time - like direction since the @xmath139 condition holds . around the @xmath94 horizon ,
the @xmath140 component changes its sign .
we can interpret that , as the area increases , some effective vacuum energy can be gained from the brans - dicke field .
however , as the effective vacuum energy changes its sign , the area ceases to increase and eventually decreases for an in - going observer .
we call this an _ instantaneous inflation _ driven by gravitational collapse .
we can compare the energy - momentum tensors by two sectors : @xmath152 where @xmath153 and @xmath154 are defined by equations [ eq : t_scalar ] and [ eq : t_bd ] . in figure
[ fig : lesst2 ] , we compare these two sectors . during the matter supply , the matter part and the brans - dicke part
are equally important for @xmath3 . however , in other cases , the brans - dicke part dominates . in figure
[ fig : moret ] , we can see the energy - momentum tensor components of the @xmath106 case .
one interesting observation is that @xmath3 has a negative region when the @xmath93 horizon shrinks .
the negative part of @xmath3 originates from the negative part of @xmath151 ( figure [ fig : moret2 ] ) . as @xmath24 increases
, the sign of @xmath151 is changed by @xmath155 .
this is an opposite behavior than the @xmath12 case ; in this case , @xmath151 was changed as @xmath156 .
these differences are consistent with the late time behaviors .
when the matter collapse finishes , @xmath151 dominates the total energy - momentum tensor .
hence , in the @xmath106 case , the apparent horizon is space - like ; in the @xmath12 case , the apparent horizon is time - like .
finally , we discuss the violation of weak cosmic censorship @xcite for the @xmath106 case .
the existence of a cauchy horizon that is not hidden by an apparent horizon implies the violation of weak cosmic censorship .
however , it is interesting to check whether the existence of the cauchy horizon can be maintained infinitely or there should be a singularity and an apparent horizon , as in the ( @xmath105 , @xmath90 , @xmath91 ) case . we argue that if the asymptotic location of the singularity has an upper @xmath23 coordinate relative to the location of the cauchy horizon , we can not see the singularity forever .
we briefly remark on a property of our double - null coordinates . *
eventually , our black holes should approach a schwarzschild black hole , and in this limit , we can use the kruskal coordinates @xcite so that @xmath157 where @xmath158 is the mass of the schwarzschild black hole
. then , the singularity is @xmath159 and the horizon is @xmath160 .
+ if we choose @xmath80 at the horizon , then for a fixed @xmath24 , the distance between the horizon and the singularity in terms of the @xmath23 coordinate becomes @xmath161 .
then , as @xmath24 increases , the singularity will approach the horizon in terms of the @xmath23 coordinate . *
the location of the apparent horizon @xmath93 will follow the case of a schwarzschild black hole . also , we know that the horizon of a schwarzschild black hole is only determined by the mass
. therefore , if we fix @xmath74 and @xmath73 , and hence we fix the asymptotic mass , for any @xmath7 , we will see asymptotically the same @xmath23 coordinate for the apparent horizon .
* therefore , the final location of a singularity is the same if we fix @xmath74 and @xmath73 .
this is observed in figure [ fig : varying_omega ] .
therefore , first , to find the location of the singularity , for given @xmath74 and @xmath73 , we will see a black hole of the @xmath147 case .
. however , the @xmath147 case clearly shows asymptotic time - like horizons .
the location of a singularity should have upper @xmath23 coordinate relative to that of a time - like horizon .
therefore , the @xmath147 case is useful to clarify the location of the singularity . ]
second , we compare the location of a cauchy horizon for the @xmath105 case with the same @xmath74 and same @xmath73 .
if the location of a cauchy horizon in @xmath105 has smaller @xmath23 than the location of a singularity in @xmath147 , this confirms that the existence of a cauchy horizon can be maintained infinitely .
contour diagrams of the radial function @xmath28 .
@xmath91 , @xmath162 , and @xmath147 and @xmath163 .
spacing is @xmath53 . ]
contour diagrams of the radial function @xmath28 .
@xmath90 , @xmath164 , and @xmath165 and @xmath163 .
spacing is @xmath53 . ] in figures [ fig : varying_a ] and [ fig : varying_vf ] , we present contour diagrams with various initial conditions . in figure [
fig : varying_a ] , we fixed @xmath91 and changed @xmath166 . also , we compared @xmath147 and @xmath105 . in the cases of @xmath167 and @xmath168 ,
the locations of the singularity between @xmath147 and @xmath105 converge to the same location .
this confirms our previous remark on the double - null coordinates .
therefore , in the cases of @xmath169 and @xmath170 , we can check that the singularity will not appear below the cauchy horizon , and hence the cauchy horizon will be maintained infinitely . in figure
[ fig : varying_vf ] , we fixed @xmath90 and changed @xmath164 . also , we compared @xmath147 and @xmath105 .
as in the previous paragraph , we can see that @xmath171 and @xmath172 cases allow infinite extension of the cauchy horizon .
classification of solutions . ] we study responses of the brans - dicke field due to gravitational collapses of scalar field pulses using numerical simulations . in this paper , we used the double - null coordinates to implement the numerical simulations
. if the absolute value of the brans - dicke coupling constant @xmath10 is sufficiently large , the responses of the brans - dicke field are negligible and eventually an einstein black hole will be formed . however ,
if @xmath7 is sufficiently small such that the brans - dicke field becomes sufficiently dynamic , we can see fully dynamical back - reactions of the brans - dicke field .
if we supply a scalar field pulse , it will asymptotically form a black hole via dynamical interactions of the brans - dicke field .
hence , we observed the responses of the brans - dicke field by two different regions ( figure [ fig : domain ] ) .
first , we observed the late time behaviors _ after _ the gravitational collapse , which includes formations of a singularity and an apparent horizon .
second , we observed the fully dynamical behaviors _ during _ the gravitational collapse and viewed the energy - momentum tensor components . for the late time behaviors ,
if @xmath7 is greater ( or smaller ) than @xmath0 , the brans - dicke field decreases ( or increases ) during the gravitational collapse ( figure [ fig : sb ] ) .
since the brans - dicke field should be relaxed to the asymptotic value with the elapse of time , the final apparent horizon becomes time - like ( or space - like ) ( figure [ fig : varying_omega ] ) .
for the dynamical behaviors , we observed the energy - momentum tensors around @xmath111 .
if @xmath12 , the @xmath2 component can be negative at the outside of the black hole .
this allows an instantaneous inflating region during the gravitational collapse ( figure [ fig : lesst ] ) . if @xmath106 , the oscillation of the @xmath3 component allows the apparent horizon to shrink ( figure [ fig : moret ] ) . in this case
, we could find a combination of initial conditions that allows a violation of weak cosmic censorship ( figures [ fig : varying_a ] and [ fig : varying_vf ] ) . in figure
[ fig : classification ] , we summarize and classify all solutions in this paper . *
if @xmath10 is sufficiently large , the @xmath93 horizon is space - like . * if @xmath101 is sufficiently small , then the @xmath93 horizon can be time - like after the matter collapse . *
if @xmath12 , there can be @xmath94 horizons .
after the matter collapse , the @xmath93 horizons are time - like . *
if @xmath106 , we can find an initial condition that the @xmath93 horizon shrinks to the singularity and eventually a space - like @xmath93 horizon and a singularity appears .
we can see a cauchy horizon . *
if @xmath106 , we also find an initial condition that the @xmath93 horizon shrinks but not to a singularity
. then , we see an oscillation of the @xmath93 horizon . * in @xmath106 case
, we can find an initial condition that the cauchy horizon can be maintained infinitely .
finally , we remark on some physical implications of our results .
first , in the @xmath12 limit , we could see a violation of the null energy condition for the @xmath2 component .
this condition is a necessary condition to induce a bubble universe using a false vacuum bubble @xcite .
it is necessary to check further calculations to ascertain whether this is in fact useful ; but one lesson is that , in brans - dicke theory or similar type of theory , we may apply the back - reaction of a gravitational collapse to induce the negative @xmath2 part .
second , in the @xmath106 limit , we could see a violation of weak cosmic censorship . even though the cauchy horizon extends infinitely
, it is reasonable to speculate that a schwarzschild black hole will be formed beyond the cauchy horizon .
however , it is also true that we can not determine what happens beyond the cauchy horizon .
these situations were not observed using numerical simulations by previous authors @xcite .
therefore , there should be further studies for this clear counterexample of weak cosmic censorship .
the authors would like to thank ewan stewart , alex nielsen , gungwon kang , and sungwook hong for discussions and encouragement .
this work was supported by korea research foundation grants ( krf-313 - 2007-c00164 , krf-341 - 2007-c00010 ) funded by the korean government ( moehrd ) and bk21 .
for observational tests , it is known that the value of @xmath7 should be greater than @xmath173 @xcite . however , in various physical situations , small @xmath7 parameters can be allowed . even though the small @xmath7 is not for our universe ,
if small @xmath7 is allowed in fundamental theory and may be realizable in somewhere of multiverse , and if such small value of @xmath7 has some implications , e.g. , violation of unitarity or cosmic censorship , the study of responses of the brans - dicke field for small @xmath7 will have theoretical importance .
one example is the dilaton gravity which has the effective action as the following form @xcite : @xmath174 where @xmath66 is the space dimensions , @xmath175 is the length scale of string units , @xmath5 is the ricci scalar , and @xmath15 is the dilaton field . if we define @xmath6 as @xmath176 where @xmath177 is the @xmath178 dimensional gravitation constant , then we obtain the brans - dicke theory with @xmath116 limit .
if there are higher loop corrections from string theory , there will be other coupling terms of @xmath6 and hence the brans - dicke theory should be modified .
hence , the correspondence between the dilaton gravity and the brans - dicke theory is only for weak coupling limit ( @xmath179 ) .
however , it is reasonable to think that brans - dicke theory with @xmath11 limit is a good toy model to study phenomena of dilaton gravity . in the first model of randall and sundrum @xcite
, they introduced two branes to explain the hierarchy problem .
because of the warp factor between two branes , we obtain a positive tension brane and a negative tension brane in an anti de sitter background . according to garriga and tanaka @xcite
, each branes can be described by brans - dicke theory in the weak field limit with the @xmath7 parameter @xmath180 where @xmath181 is the location of the negative tension brane along the fifth dimension , @xmath182 is the length scale of the anti de sitter space , and the sign @xmath183 denotes the sign of the tension . to explain the hierarchy problem , we require @xmath184
. then we obtain sufficiently large @xmath7 on the positive tension brane and @xmath185 on the negative tension brane .
however , in principle , @xmath186 can be chosen arbitrarily and hence one may guess that various @xmath7 near @xmath187 may be allowed by models from brane world . for more discussions , see @xcite .
in this appendix , we report on convergence and consistency tests for our simulations .
we used @xmath11 , @xmath90 , and @xmath91 case . for convergence , we compared finer simulations : @xmath188 , @xmath189 , and @xmath190 times finer . in figure
[ fig : convergence ] , we see that the difference between the @xmath188 and @xmath189 times finer cases is @xmath108 times the difference between the @xmath189 and @xmath190 times finer cases , and thus our simulation converges to second order . for @xmath191 and @xmath192 ,
the curves appear to diverge , since there is a singularity .
however , even before the singularity , the error is @xmath193 .
plots of errors with different step sizes . here , we plot @xmath194 and @xmath195 along a few constant @xmath23 lines , where @xmath196 is calculated in an @xmath197 times finer simulation than @xmath198 . since
two curves are folded for each @xmath23 slice , we confirm that our simulation converges to second order . ]
we also compared constraint equations ( equations ( [ eq : e1 ] ) and ( [ eq : e2 ] ) ) using @xmath201 figures [ fig : constraint_u ] and [ fig : constraint_v ] show the constraint equations . around the singularity ,
the constraint equations increase , but still the value is less than @xmath202 . for figure
[ fig : constraint_v ] , the constraint equation initially appears to be large , because initially both the numerator and denominator , in equation ( [ eq : constraint_v ] ) , are too small ( less than @xmath203 ) .
therefore , these plots show that our simulations hold the constraint equations sufficiently .
m. a. scheel , s. l. shapiro and s. a. teukolsky , phys .
d * 51 * , 4208 ( 1995 ) [ arxiv : gr - qc/9411025 ] ; + m. a. scheel , s. l. shapiro and s. a. teukolsky , phys . rev .
d * 51 * , 4236 ( 1995 ) [ arxiv : gr - qc/9411026 ] .
t. piran and a. strominger , phys .
d * 48 * , 4729 ( 1993 ) [ arxiv : hep - th/9304148 ] ; + r. parentani and t. piran , phys .
lett * 73 * , 2805 ( 1994 ) [ arxiv : hep - th/9405007 ] ; + s. ayal and t. piran , phys .
d * 56 * , 4768 ( 1997 ) [ arxiv : gr - qc/9704027 ] ; + s. hod and t. piran , phys .
lett * 81 * , 1554 ( 1998 ) [ arxiv : gr - qc/9803004 ] ; + s. hod and t. piran , gen .
grav . *
30 * , 1555 ( 1998 ) [ arxiv : gr - qc/9902008 ] ; + e. sorkin and t. piran , phys .
d * 63 * , 084006 ( 2001 ) [ arxiv : gr - qc/0009095 ] ; + e. sorkin and t. piran , phys .
d * 63 * , 124024 ( 2001 ) [ arxiv : gr - qc/0103090 ] ; + y. oren and t. piran , phys .
d * 68 * , 044013 ( 2003 ) [ arxiv : gr - qc/0306078 ] ; + j. hansen , a. khokhlov and i. novikov , phys . rev .
d * 71 * , 064013 ( 2005 ) [ arxiv : gr - qc/0501015 ] ; + a. doroshkevich , j. hansen , i. novikov and a. shatskiy , arxiv:0812.0702 [ gr - qc ] ; + a. doroshkevich , j. hansen , d. novikov , i. novikov and a. shatskiy , arxiv:0908.1300 [ gr - qc ] .
a. ashtekar and b. krishnan , living rev .
* 7 * , 10 ( 2004 ) [ arxiv : gr - qc/0407042 ] ; + s. a. hayward , arxiv : gr - qc/9303006 ; + a. b. nielsen and d. yeom , int .
j. mod .
a * 24 * , 5261 ( 2009 ) [ arxiv:0804.4435 [ gr - qc ] ] . | we study responses of the brans - dicke field due to gravitational collapses of scalar field pulses using numerical simulations .
double - null formalism is employed to implement the numerical simulations .
if we supply a scalar field pulse , it will asymptotically form a black hole via dynamical interactions of the brans - dicke field .
hence , we can observe the responses of the brans - dicke field by two different regions .
first , we observe the late time behaviors _ after _ the gravitational collapse , which include formations of a singularity and an apparent horizon .
second , we observe the fully dynamical behaviors _ during _ the gravitational collapse and view the energy - momentum tensor components . for the late time behaviors ,
if the brans - dicke coupling is greater ( or smaller ) than @xmath0 , the brans - dicke field decreases ( or increases ) during the gravitational collapse .
since the brans - dicke field should be relaxed to the asymptotic value with the elapse of time , the final apparent horizon becomes time - like ( or space - like ) .
for the dynamical behaviors , we observed the energy - momentum tensors around @xmath1 . if the brans - dicke coupling is greater than @xmath0 , the @xmath2 component can be negative at the outside of the black hole .
this can allow an instantaneous inflating region during the gravitational collapse .
if the brans - dicke coupling is less than @xmath0 , the oscillation of the @xmath3 component allows the apparent horizon to shrink .
this allows a combination that violates weak cosmic censorship .
finally , we discuss the implications of the violation of the null energy condition and weak cosmic censorship . |
while cosmology appears to be converging on a `` concordance model '' that describes the universe as inflationary and isotropic , there remain unexplained anomalies in the cmb data , and other models are not yet ruled out .
the _ wmap _ data provide some of the most accurate measurements yet of the cosmic microwave background and contribute to high accuracy determinations of cosmological parameters @xcite .
however , there are several studies that show that at large scales , the cmb is not statistically isotropic and gaussian , as predicted by inflation theory @xcite . in @xcite , we examined a particular set of anisotropic cosmological models : the bianchi type vii@xmath0 class of spatially homogeneous generalizations of friedmann universes that include small vorticity ( universal rotation ) and shear ( differential expansion ) components .
surprisingly , we found evidence that one of these models correlates with the cmb sky and that subtracting this component resolves several of these observed anomalies .
the models used in that study were derived by @xcite and include no cosmological constant component in the total energy density .
the best - fit model found by that study required @xmath2 , implying , @xmath3 , i.e. , a significantly negatively curved universe .
@xcite subsequently considered flat bianchi models and sought explicitly to resolve the problems of the low quadrupole and the low-@xmath4 alignments using a statistic constructed to achieve that purpose .
their analysis does indeed find a model that fixes these anomalies , but it remains statistically insignificant as a detection .
our original result has the benefit of being a detection that is based entirely on a simple least - squares fit of the bianchi template to the data , yet we find that it serendipitously resolves several anomalies without that requirement having been built - in to the search algorithm .
the importance of our result , therefore , lies in the fact that it resolves the problems of the low quadrupole , low-@xmath4 alignments , power asymmetry , and non - gaussian cold spot
. however , both analyses neglect the fact that the existing bianchi solutions include no dark energy component .
furthermore , the best - fit results require an energy content that is consistent neither with other astronomical observations nor with the cmb itself . in this work ,
we present a modification of the barrow et al .
bianchi type vii@xmath0 solution so that it includes a cosmological constant term in the evolution .
we discuss the impact of that term on the structure of the resulting cmb anisotropy pattern , particularly the degeneracy that is introduced in the model space by the addition of @xmath5 .
we then discuss the viability of @xmath6 models that are morphologically identical to our best - fit @xmath7 models by considering the constraints imposed by different measurements of the cosmological parameters .
first we shall generalize the equations in @xcite to include a cosmological constant @xmath6 .
the basic assumption is that the type vii@xmath0 universe is close to frw .
we start with the equations of motion using expansion - normalized variables ( see , e.g. , * ? ? ?
the equations describing the evolution of bianchi type vii@xmath0 universes with a tilting perfect fluid can be found in @xcite and @xcite . in our case
, we consider a tilting perfect fluid with @xmath9 ( dust ) and a cosmological constant . the equations in @xcite can easily be generalized to include a cosmological constant by adding a matter term @xmath10 wherever there is a matter term @xmath11 .
furthermore , we assume that the universe is close to a frw at all times and that the tilt velocity of the dust fluid is small ; hence we assume : @xmath12 here , @xmath13 is the expansion - normalized shear scalar , @xmath14 is a complex curvature variable , and @xmath15 , @xmath16 are the tilt components ( see @xcite for details ) . the curvature variables
@xmath17 and @xmath18 need not be small , and they are related to the group parameter @xmath19 as follows : @xmath20 we also assume that the parameter @xmath19 is not too small : @xmath21 .
( as @xmath22 , the assumptions made for this derivation break down and a qualitatively different solution for type vii@xmath23 is needed .
see @xcite . )
given the above assumptions , the deceleration parameter is @xmath24 and the equation of motion for the hubble scalar is @xmath25 where the prime indicates the derivative with respect to the dimensionless time @xmath26 defined by @xmath27 , and @xmath28 is the cosmological time . at the lowest order ,
the tilt components that are related to the vorticity are the components @xmath29 ( in the notation of * ? ? ?
these induce non - zero shear components @xmath30 via the linear constraint equation , which in the frw limit reduces to : @xmath31 moreover , close to frw the curvature @xmath32 and matter densities @xmath33 and @xmath34 are related via the friedmann equation @xmath35 we also define @xmath36 to be @xmath37 eq .
( [ constraint ] ) now reduces to barrow et al.s eqs.(4.6 ) and ( 4.7 ) by dropping the subscript @xmath38 and replacing @xmath39 with @xmath33 .
the equations of motion can now be solved . by choosing @xmath40 , @xmath41
, we can relate @xmath42 to @xmath26 and the time variable @xmath43 in @xcite to the redshift @xmath42 : @xmath44 the hubble scalar , @xmath45 , @xmath33 , and @xmath34 can then be written @xmath46 @xmath47^{\frac 12},\label{eq : h_of_z}\ ] ] @xmath48 @xmath49 @xmath50 where @xmath51 are the present values of the expansion - normalized @xmath6 , curvature and ordinary matter ( dust ) and obey the constraint : @xmath52 we also note that @xmath53 implies negative curvature .
the shear components @xmath54 and @xmath55 , and the remaining curvature variable are given by @xmath56 @xmath57 @xmath58 conformal time becomes @xmath59 which can be used in the barrow expressions for @xmath60 and @xmath61 .
the integrals of barrow et al .
equations ( 4.12 ) and ( 4.13 ) become @xmath62 @xmath63 with @xmath64 , where @xmath65 is the redshift at photon emission . in calculating the constant @xmath66
, the density @xmath67 should now be @xmath68 , i.e. the matter density only .
the following assumptions should be kept in mind regarding this solution : * the tilt velocity is small . in a dust- or @xmath6-dominated bianchi type vii@xmath0 universe
, the tilt asymptotically tends to zero at late times . using the above relations
, it can also be shown that for the case of interest , it remains small back to the last scattering surface as well . *
the universe is close to frw at all times .
this assumption is required to make the problem tractable . in a @xmath6-dominated universe ,
the type vii@xmath0 models become close to frw . for the cases in question
, plugging the numbers into the above relations shows that this assumption still holds at the surface of last scattering .
( at very early times , however , this is not the case . )
it is likely that models for which this assumption does not hold at last scattering would then have greater shear effects , but the details would require a more complicated derivation .
* @xcite only consider the contribution from the vorticity that directly involves the shear variables @xmath54 and @xmath55 .
the ( expansion - normalized ) shear scalar , @xmath13 , is given by @xmath69 and involves therefore all of the shear components . in @xcite and here , the other shear components have simply been ignored in the calculation of @xmath70 . considering additional shear degrees of freedom
is in principle possible but in practice difficult .
the effect of such additions is one of the largest unknowns in this analysis . in the presence of the cosmological constant , a universe that is very nearly flat now and that was very nearly flat at the time of last scattering can become significantly negatively curved at intermediate redshifts .
we can see this from eqtn .
[ eq : sigbjorn12 ] above .
figure [ fig : k_z ] shows how @xmath71 evolves for different values of @xmath72 and @xmath68 . at high redshift ,
the matter dominates and the curvature vanishes , and at low redshift , @xmath5 can dominate , depending on its exact value . at intermediate redshifts ,
however , the universe may become more negatively curved .
for a currently almost flat universe with parameters close to the observed values of @xmath73 and @xmath74 , the curvature is at most only a few percent ( black curve ) . only for very small matter densities and very large dark energy densities
does the curvature become large at intermediate redshifts . to reproduce
the observed asymmetry with these @xmath75 models requires either a large current negative curvature ( pink and blue dot - dashed curves ) or a very small matter density ( green dashed curve ) .
the addition of a @xmath6 component adds a third parameter to the bianchi parameter space , but there is a degeneracy to the resulting morphology of the bianchi - induced pattern .
the same structure in the induced anisotropy pattern can be reproduced by different parameter combinations .
the addition of dark energy results in a tighter spiral for a given value of @xmath36 , because the redshift to the surface of last scattering corresponds to a larger time difference .
the geodesics have therefore completed more rotations since recombination for a given value of @xmath36 when @xmath1 .
secondly , the focusing of the spiral depends on the curvature between the observer and the last scattering surface .
significant focusing can arise from a universe that is very open now , or from a universe where @xmath6 is large enough to give negative curvature at intermediate redshifts ( as shown in fig .
[ fig : k_z ] . ) three combinations of parameters , for example , can reproduce the same apparent structure on the sky : 1 .
[ item : one ] @xmath76 i.e. , no @xmath6 and a matter content half critical , giving a large current negative curvature ( the model found in * ? ? ?
* ) ; 2 .
[ item : two ] @xmath77 i.e. , approximately the observed matter content ( baryon plus dark ) , with some @xmath6 but still a large current negative curvature ; 3 .
[ item : three ] @xmath78 i.e. , a much smaller than observed matter content , but a nearly flat current curvature ( @xmath79 ) ; the large @xmath6 causes a large negative curvature at intermediate redshifts .
the real difference between these models is the amplitude : ( [ item : three ] ) has an amplitude @xmath80 of ( [ item : one ] ) , and ( [ item : two ] ) has an amplitude of @xmath81 _ times _ that of ( 1 ) .
the shear and vorticity values obtained from fitting these templates to the data would change accordingly .
effectively , this implies that a given structure on the sky characterized by the amount of geodesic focusing and the number of spiral turns is the same for all models along a line in the three dimensional parameter space of @xmath82 .
the degeneracy in the templates is only broken by the amplitude of the variation .
but @xmath83 is what we measure by fitting a template to the sky , so we can not distinguish among the degenerate models without an independent measurement of the shear .
it is not straightforward to calculate where such lines of degeneracy lie .
instead , we determine this empirically by generating a grid of models and comparing them to the previously determined best - fit model .
we now address the question of whether the degeneracy of the bianchi type vii@xmath19 models with @xmath84 described in [ sec : models ] gives us enough freedom to define a bianchi model that is morphologically identical to the best - fit model in @xcite and that is also consistent with estimates of the cosmological parameters .
@xcite describe the limitations of constraining cosmological parameters with cmb data alone .
in particular , there is a geometric degeneracy in the curvature and dark energy density ( or equivalently , the matter density . )
degenerate models have : 1 ) the same values of the physical baryon density , @xmath85 , and cold dark matter density , @xmath86 ; 2 ) the same primordial fluctuation spectrum ; and 3 ) the same value of the parameter @xmath87 & \text{if } \omega_k > 0 , \\ \sqrt{\omega_k}y & \text{if } \omega_k = 0 , \\ \sin[\sqrt{\omega_k}y ] & \text{if } \omega_k < 0 , \end{cases}\ ] ] where @xmath88 ( where @xmath89 is the scale factor at recombination ) .
these models will have almost the same power spectra .
contours of constant @xmath90 in model space @xmath91 versus @xmath5 are shown in figure [ fig : degen ] , with red highlighting the curve intersecting the best - fit , flat , _ wmap_-only parameters ( * ? ? ?
* table 1 ) .
figure [ fig : degen_cls ] shows power spectra for those degenerate models ( calculated using cmbfast .
) when normalized , the spectra are identical at the acoustic peaks , and the degeneracy only fails at large angular scales . in the three dimensional parameter space of @xmath92 ,
the geometric degeneracy forms a surface , while the bianchi models with identical structure form a line .
the two need not intersect , but in the case of the observed _ wmap _ cmb - only power spectrum of @xcite and best - fit bianchi model of @xcite , they do .
there is a bianchi model that has the identical structure to the best - fit @xmath7 model and lies on the _ wmap _ degeneracy curve .
it has parameters @xmath93 . but
how viable is this region of parameter space ?
as shown in figure [ fig : degen_cls ] , the geometric degeneracy is broken at large angular scales .
models with high @xmath94 and low @xmath95 predict too much large scale power .
_ wmap _ data alone place relatively loose constraints on @xmath95 , but even these rule out such low values as required for the bianchi models .
( the power spectrum of the sky corrected for the bianchi component has less large - scale power overall @xcite than the uncorrected power shown in that figure , so the problem will only become worse . ) to quantify these limits , we examine the posterior likelihood from the _ wmap _ data alone using the cosmomc code of @xcite , which implements a monte carlo markov chain method . for a simple look to compare the _ wmap _ constraints to the bianchi degeneracy
, we explore the five - parameter space of : @xmath96 ( the ratio of the sound horizon to the angular diameter distance ) , @xmath43 ( optical depth ) , @xmath91 ( curvature ) , @xmath97 ( spectral index of scalar perturbations ) , and @xmath98 ( amplitude of scalar perturbations ) , with fixed parameters @xmath99 ( the physical baryon density ) , @xmath100 ( the physical dark matter density ) , @xmath101 ( neutrino fraction of dark matter density ) , @xmath102 ( dark energy equation of state ) , @xmath103 ( ratio of tensor to scalar fluctuations ) .
other parameters such as @xmath104 , @xmath94 , and @xmath95 are derived from this set .
( the free parameters have only such priors as defined by ranges that are much broader than any realistic uncertainty . ) the resulting constraints in the @xmath105 plane ( marginalizing over the other parameters ) are shown in figure [ fig : om_ol_space ] in red . all bianchi models with the structure of the best - fit model lie well outside the 95% confidence region from the _ wmap _ data alone . allowing the physical matter densities to vary as well but including other datasets ,
the constraints ( in magenta ) are even tighter .
@xcite give @xmath106 from _ wmap _ data alone , which tightens to @xmath107 adding data from the sloan digital sky survey ( sdss ) .
see also @xcite for cmb+2dfgrs results .
figure [ fig : degen_cls ] shows that the geometric degeneracy breaks at large angular scales , and the models with high @xmath5 have too much large scale power .
but there is an additional degeneracy if we allow the optical depth , @xmath43 , to vary .
adjusting @xmath43 and @xmath98 such that @xmath108 remains constant has exactly the effect we need of modifying only the large scale power while leaving the peak heights unchanged @xcite .
figure [ fig : cls_tau ] shows how this works , but that the effect is not large enough .
even a @xmath43 of zero does not bring the power at large angular scales within range of the data , and that model is inconsistent with the large - scale peak in the te spectrum .
one can also ask if the addition of the bianchi component can affect the variance of the large scale te cross - power , resulting in an incorrect estimate of @xmath43 due to a chance alignment of the polarization signal with the bianchi structure .
simulations with no reionization and with an added bianchi temperature pattern ( the same location and shear amplitude as our best - fit model ) show that this is not the case .
the cross - power remains flat at low @xmath4 with the expected variance .
note that @xcite find different values of @xmath43 derived from fitting temperature data in the northern and southern hemispheres as defined in the frame of reference that maximizes the power asymmetry ( see @xcite . )
the additional large - scale structure in the south that the bianchi template reproduces could cause this effect . if a polarization signal were also produced in a bianchi geometry and correlated with the temperature anisotropy , it would influence the measured optical depth from the low-@xmath4 te peak as well .
the degeneracy contours in figure [ fig : degen ] are plotted for constant values of @xmath109 and @xmath99 ( from * ? ? ? * ) . for different points in the @xmath110 space , @xmath68 changes and therefore the value of @xmath111 changes , increasing with increasing @xmath72 along the degeneracy curve . for models of high @xmath72 that lie on the geometric degeneracy curve
, @xmath104 reaches values over @xmath112 .
this is ruled out at high significance by the _ wmap _ data itself ; @xcite give @xmath113 .
independent determinations of hubble s constant also rule out these values , e.g. , the hst key project @xcite value of @xmath114 .
see , e.g. , ( * ? ? ?
* table 3 ) for other estimates of the hubble constant .
high - redshift supernovae observations can break the geometric degeneracy by independently placing constraints that are nearly perpendicular to the _ wmap _ constraints in @xmath115 space ( see figure [ fig : om_ol_space ] . )
the degeneracy curve for the best - fit bianchi model compared to the _ wmap _ data lies outside the @xmath116 confidence contour determined by the supernova cosmology project ( * ? ? ?
* fit no . 6 ) .
it is worth pointing out that it is not clear whether type ia supernovae are truly standard candles out to high redshift .
the uncertainty , however , is not enough to accommodate such low values of @xmath95 .
the addition of cosmic vorticity and shear would , of course , influence the determinations of such parameters .
the sky coverage of supernova , for example , is fairly extensive outside the galactic plane region , but the sample size is small .
studies such as @xcite have ruled out significant dipole or quadrupole asymmetry in the expansion , but a test for a more complicated anisotropy structure induced by the vorticity would require far more supernova in the sample and a good sky coverage . furthermore , as the current value of the shear expansion is very small , distance measurements to relatively low red - shift ( compared to the cmb ) objects may not be sensitive enough to detect it .
this also implies that such a small current shear would not significantly alter measurements of other cosmological parameters such as the hubble constant that are dependent on relatively low - redshift observations .
one can also ask , then , how much one can vary the parameters of the bianchi model and still have a statistically significant fit
. the most interesting region on figure [ fig : om_ol_space ] is near @xmath117 , where the _ wmap _ and supernovae constraints approach the region of the best - fit bianchi model .
this model resembles the best - fit model , but since it is closer to flat than the bianchi degeneracy curve , there is less geodesic focusing . as a result , the model is not so good a fit to the data , and has a significance , compared to gaussian realizations , of only @xmath118 , compared to over @xmath119 for the model at the same @xmath5 but on the bianchi degeneracy curve .
considering that the _ wmap _ data somewhat favor a closed universe , one might ask about closed models with vorticity and shear .
these are bianchi type ix models , also discussed in @xcite . unlike the open type vii@xmath0 models ,
however , closed models exhibit neither geodesic focusing nor the spiral pattern , even in the presence of vorticity .
barrow et al .
use them to place limits on vorticity and shear simply using the quadrupole .
such closed models do not reproduce the morphology needed to explain the power asymmetry or the cold spot , and it would be impossible to claim any detection with only the quadrupole as an observable .
the cosmological constant is the simplest form of dark energy , and all observations so far remain consistent with it .
other models are not ruled out , however , and in some cases are favored .
alternatives come in many varieties , some with physical motivation , others constructed to give a particular result .
( see , e.g. , @xcite . ) here , we address the question of whether an alternative dark energy model can bring our bianchi pattern any closer to the constraints imposed by the data .
dark energy models are characterized by their equation of state , @xmath120 , where a cosmological constant @xmath6 corresponds to a model with a constant @xmath102 .
alternative theories allow @xmath121 to vary with time , as in `` quintessence '' and `` k - essence '' models .
a rather _ ad hoc _ parameterization is often used of the form @xmath122 , which allows comparison of generic dark energy models with supernova data @xcite .
the derivation in
[ sec : models ] can be generalized for any dark energy model by the appropriate substitution into equations [ eq : h_of_z ] and [ eq : lambda ] of @xmath123 instead of @xmath5 .
we then have @xmath124 where the function @xmath125 is derived from the equation of state as @xmath126 examples for extreme values of @xmath127 and @xmath128 are shown in figure [ fig : k_z_de ] .
as was shown in figure [ fig : k_z ] , we can look at the evolution of the curvature as a function of time for these models and see if any display the required negative curvature at intermediate redshift . using this parameterization , or
alternatively the parameterization @xmath129 , does not give much additional freedom , however , to create a viable model .
ultimately , the amount of curvature and therefore focusing is still largely a function of the relative densities of matter and dark energy and of the current curvature . manipulating the evolution of the dark energy equation of state simply changes the time scales
on which the transitions between the different components of the density occur .
we have so far ignored an additional issue with these models related to the stochastic component of the cmb .
the analysis in @xcite assumes that the observed cmb anisotropies consist of two independent signals : the predicted bianchi pattern , and a stochastic , statistically isotropic component .
the latter may be generated via inflation or another mechanism , but if it is statistically isotropic at the surface of last scattering , it might no longer be statistically isotropic when observed after traveling through a bianchi universe .
the power spectrum at small scales might deviate from predictions due to additional structure ( dependent on orientation relative to the preferred axis ) introduced by propagation in a bianchi metric .
we note that the asymmetry observed in @xcite and @xcite extends to @xmath130 , while the bianchi model has structure only up to @xmath131 .
geodesic focusing might cause such a power asymmetry , though it might also require that the asymmetry continue to the smallest angular scales , which is not observed .
@xcite also find that some unexplained outliers in the _ wmap _ power spectrum are associated with different hemispheres .
furthermore , the generation of the fluctuations at the last scattering surface might also be affected by the anisotropy , though in the case in question , that anisotropy remains very small ( @xmath132 ) at @xmath133 .
essentially , we expect that the bianchi models could also be constrained by the lack of deviations in the power spectrum from the best - fit _ wmap_/concordance model , although detailed predictions for such deviations are needed
. the consistency of the acoustic peak scale with the flat concordance model would be very difficult for our best - fit bianchi model to explain .
considering the success of big bang nucleosynthesis in explaining light element abundances , any model that significantly changes the physical processes at those early times must be ruled out .
several studies have thus attempted to place limits on shear expansion and rotation by examining their effect on the relative helium abundance , @xmath134 .
@xcite originally showed that such analysis can place strong constraints on shear .
the interesting exception , however , was type vii@xmath0 models , where barrow found that the cmb remained a stronger constraint .
later studies showed that including more complicated effects can reverse the trend of @xmath134 with the shear ( see , e.g. , @xcite ) .
@xcite revisited the issue and showed that in some cases , even extremely anisotropic models may still have the observed helium abundance .
bianchi type vii@xmath0 models , however , have not been treated in detail since the result of @xcite . after demonstrating for several cases , not including vii@xmath0 , that @xmath134 increases for anisotropic models , @xcite _ conjecture _ that the same trend applies for _ all _ anisotropic models , but they admit that this is difficult to prove . in short ,
it is possible that such bianchi models are strongly ruled out by bbn and the observed helium fraction .
but this has not been definitively proved .
we have presented solutions for bianchi type vii@xmath0 type universes that include a dark energy term and examined how their morphological properties change over the parameter space .
the addition of dark energy adds a degeneracy such that different combinations of the three parameters @xmath135 can lead to the same observed structure as in the best - fit model of @xcite .
a template can be constructed that has the identical structure of that best - fit model and also falls on the geometric degeneracy curve for the parameters as measured by _
wmap _ data .
this model , however , lies well outside the 95% confidence region in @xmath115 space for _ wmap _ data , ruled out by the over - prediction of large scale power .
it also lies outside the 95% likelihood contours from the supernova cosmology project , and is further inconsistent with the measurement of the hubble constant from the hubble key project .
bianchi models that are more consistent with these other measurements are no longer good fits to the _ wmap _ large scale structure .
one of the most difficult problems for these models is to account for the acoustic peak structure .
the anisotropy at early times might influence the nature of the fluctuations at last scattering , and the geometry could affect the power spectrum on small angular scales due to the geodesic focusing between last scattering and the observer .
detailed predictions for these effects would be needed , but it is difficult to envision such an anisotropic scenario that happened to reproduce the observed acoustic peak structure , mimicking the concordance cosmology so well .
there is currently no prediction for the cmb polarization anisotropy in a bianchi universe , but such a geometry - induced signal could provide an additional test of these models . if the preferred direction indicated in the temperature data is also reflected in the full sky polarization data ( expected from further _
wmap _ data releases and , eventually , from planck ) , there will be even more motivation to consider non - standard models .
we have shown that our best - fit bianchi type vii@xmath0 model is not compatible with measured cosmological parameters , despite the additional freedom from adding dark energy .
it is worth reiterating , however , that the serendipitous discovery of a theoretically derived template that correlates well with the data also happens to resolve several anomalies that can not be explained in the standard picture .
these particular models may not be viable , but lacking any plausible scenario for systematics or foregrounds to be the source of the anomalies , non - standard models that reproduce a similar morphology merit continued interest .
we are grateful to s. d. m. white for useful discussions and suggestions .
sh was funded by a killam postdoctoral fellowship .
we acknowledge use of the healpix software @xcite and analysis package for deriving some results in this paper .
we also acknowledge use of the legacy archive for microwave background data analysis ( lambda ) . | we generalize the predictions for the cmb anisotropy patterns arising in bianchi type vii@xmath0 universes to include a dark energy component .
we consider these models in light of the result of @xcite in which a correlation was found on large angular scales between the _ wmap _ data and the anisotropy structure in a low density bianchi universe .
we find that by including a term @xmath1 , the same best - fit anisotropy pattern is reproduced by several combinations of cosmological parameters .
this sub - set of models can then be further constrained by current observations that limit the values of various cosmological parameters . in particular
, we consider the so - called geometric degeneracy in these parameters imposed by the peak structure of the _ wmap _ data itself . apparently , despite the additional freedom allowed by the dark energy component ,
the modified bianchi models are ruled out at high significance . |
starting more than a decade ago considerable observational evidence has been obtained indicating that planets are able to exist in stellar binary ( and higher order ) systems ; see results and discussions by , e.g. , @xcite , @xcite , and @xcite .
these observations are in line with the empirical finding that binary ( and higher order ) systems occur in high frequency in the local galactic neighborhood @xcite . for example , @xcite presented results of a detailed analysis of companions to solar - type stars , based on a sample size of 454 , and concluded that the overall fractions of double and triple systems are about 33% and 8% , respectively , if all confirmed stellar and brown dwarf companions are accounted for .
updated results were meanwhile given by @xcite .
this study shows that 57 exoplanet host stars are identified having a stellar companion .
the fairly frequent occurrence of planets in binary systems is furthermore consistent with the presence of debris disks in a considerably large number of main - sequence star binary systems ( e.g. , * ? ? ?
* ) . in principle , as discussed by @xcite , planets in binary systems can be identified through two different venues : first , binaries or multiple star systems can be surveyed for the presence of planets by utilizing the established detection methods .
second , stars with detected planets can be scrutinized afterward to check if they possess one or more widely separated stellar companion(s ) ; in this case , the planet(s ) will also be categorized as belonging to a binary ( or higher order ) system . from the view point of orbital mechanics , there are two different kinds of possible orbits ( notwithstanding positions near the lagrangian points l@xmath0 and l@xmath1 ) for planets in binary systems : s - type and p - type orbits @xcite .
a p - type orbit is given when the planet orbits both binary components , whereas in case of an s - type orbit the planet orbits only one of the binary components with the second component behaving as a perturbator .
@xcite presented a list of 15 planet - bearing binary systems with all planets in s - type orbits .
they constitute mostly wide binaries with separation distances of up to @xmath26400 au ; however , smaller separation distances on the order of 20 au or less have also been identified . in the meantime , systems with planets in p - type orbits have also been identified .
arguably , the most prominent case is kepler-16 , as reported by @xcite and previously suggested by @xcite , containing a saturnian mass circumbinary planet . @xcite
have subsequently studied this system regarding the possibility of habitable exoplanets and habitable exomoons .
recently , a transiting circumbinary multiplanet system , i.e. , kepler-47 , has also been identified @xcite . there is a significant body of literature devoted to the study of habitability / h@xmath3o / n@xmath3 atmosphere ( see sect . 2 for details ) .
more sophisticated approaches to habitability have been given in the meantime taking into account additional aspects , such as the planet s size and mass , atmospheric structure and composition , magnetic field , geodynamic properties , ionizing stellar uv and x - ray fluxes , and tidal locking ( if existing ) ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . additionally ,
as pointed out by @xcite , planets with sufficiently thick atmospheres may remain habitable even when temporarily absent from their hzs due to orbits of considerable ellipticity .
this possibility is disregarded in the following as well , as planets will be required to stay permanently in the chz , ghz , or ehz ( see sect . 2 for definitions ) , as applicable , to be considered habitable .
] in binary systems as well as in multiplanetary systems , which often also encompass stellar evolutionary considerations .
examples include the work by @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
an important aspect that has received increased recognition in the literature is that in order for habitability to exist and to be maintained , a joint constraint that includes both orbital stability and a habitable environment for a system planet through the stellar radiative energy fluxes needs to be met .
in the framework of this paper , the zone related to this latter requirement will subsequently be referred to as _ radiative habitable zone _ ( rhz ) , which constitutes a necessary , though often insufficient , condition for the existence of circumstellar habitability . previous work , mostly concentrated on the existence of habitability in single star multi - planetary systems , rendered the publication of detailed stability catalogs " for the habitability zones of extrasolar planetary systems ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
for example , @xcite quantified the dynamical habitability of 85 planetary systems by considering the perturbing influence of giant planets beyond the traditional hill sphere for close encounters with the theoretical terrestrial planets .
they concluded that a significant fraction of the identified extrasolar planetary systems are unable to harbor habitable terrestrial planets .
a statistical study on the stability of earth - mass planets orbiting solar - mass stars in presence of stellar companions focusing on both the statistical properties of ejection times and the general prospects of planetary habitability was given by @xcite .
additional work providing stability assessments for various observed extrasolar planetary systems based on detailed stability maps was given by @xcite .
@xcite explored the orbital stability or planets in double - planet systems for binaries by supplying an analytic framework based on secular perturbation theory ; they also provided dynamical classification categories .
additional stability analyses to assess the habitability of planetary systems based on detailed numerical simulations were given by @xcite and @xcite ; note that the study of @xcite also dealt with a limited cases of planets in double star systems in orbit either around one stellar component ( s - type ) or around both components ( p - type ) . the study of planetary dynamics and habitable planet formation has meanwhile been described by , e.g. , @xcite and @xcite .
they show that earth - mass planets are , in principle , able to form in stellar binary systems , although many details of the relevant processes are not fully understood .
the overarching conclusion of those investigations is that habitable planets in stellar binary ( and , as anticipated , in higher - order systems ) are , in general , possible , which is a stark motivation for providing a comprehensive study of s - type and p - type habitability in binary systems .
the approach adopted in this study will be entirely analytic .
specifically , it will consider both s - type and p - type habitable orbits in the view of the joint constraint including orbital stability and a habitable region for a system planet through the appropriate amount of the stellar radiative energy fluxes . in an earlier study
, @xcite focused on s - type habitability in binary systems taking into account both circular and elliptical orbits for the stellar binary components ; this latter aspect is however beyond the scope of the present work as we solely focus on systems in circular orbits .
numerical studies for p - type habitable environments with applications to kepler-16 , kepler-34 , kepler-35 and kepler-47 have been given by @xcite .
our paper is structured as follows : in sect . 2
, we comment on the adopted main - sequence star parameters and single star habitability . in sect . 3
, we introduce our theoretical approach suitable for stellar systems of the order of @xmath4 , although our focus will be on binary systems . in this regard
both s - type and p - type orbits will be examined , and detailed mathematical criteria for the existence of s - type and p - type rhzs will be derived . in sect . 4 , we consider the additional constraint of planetary orbital stability for the establishment of circumstellar habitability .
applications regarding s - type and p - type systems are given in sect .
5 , whereas the habitability classifications s , p , st , and pt are introduced in sect .
section 7 conveys our summary and conclusions .
in this study , s - type and p - type habitability is investigated mostly pertaining to standard ( i.e. , theoretical ) main - sequence stars .
the adopted stellar parameters , which are the stellar effective temperatures @xmath5 , the stellar radii @xmath6 ( which together allow to define the stellar luminosities @xmath7 ) , and the stellar masses @xmath8 are mainly based on the work by @xcite ( see his table b.1 ) that assumes detailed photospheric spectral analyses . for stellar spectral types with no data available , the missing data were computed by employing biparabolic interpolation .
the exception , however , are data for stars of spectral type k5 v and below . in this regard we relied on the results from the spectral models of r. l. kurucz and collaborators .
they took into account hundreds of millions of spectral lines for a large set of atoms and molecules ; see @xcite and @xcite for details .
the effective temperatures implied by these models are in close similarity to those given by @xcite for most types of stars ; however , @xcite reports consistently higher effective temperatures for stars of spectral type late - k and m ; for the latter , the difference amounts to nearly 300 k. table 1 depicts the stellar parameters adopted for the present work .
an alternative approach expected to provide very similar results for either the stellar luminosity or the stellar mass ( with the other parameter taken as fixed ) is the employment of a mass luminosity relationships applicable to main - sequence stars .
the work by @xcite , as well as data from subsequent studies , yield @xmath9 with @xmath10 and @xmath11 for @xmath12 and @xmath13 and @xmath14 for @xmath15 . at the high - mass end , this relationship holds until about @xmath16 .
it also becomes increasingly inaccurate for low - mass m dwarfs .
fortunately , the domains of applicability for eq .
( 1 ) is consistent with most studies of binary habitability ; see , e.g. , @xcite as example .
next we focus on single star habitability , i.e. , the evaluation of various limits of habitable zones ( hzs ) , which in the solar case shall be referred to as @xmath17 .
previous work by , e.g. , @xcite distinguished between conservative ( chz ) and the generalized habitable zone ( ghz ) , which can also be evaluated for general main - sequence stars , and other types of stars as well .
for the sun , the limits of the chz are given as 0.95 and 1.37 au ( @xmath18 = 2 and 4 , respectively ) , whereas for the ghz , they are given as 0.84 and 1.67 au ( @xmath18 = 1 and 5 , respectively ) ; see table 2 .
the physical significance of the various kinds of hzs obtained by @xcite can be summarized as follows : the ghz is defined as bordered by the runaway greenhouse effect ( inner limit ) and the maximum greenhouse effect ( outer limit ) . concerning the latter it is assumed that a cloud - free co@xmath3 atmosphere is still able to provide a surface temperature of 273 k. the inner limit of the chz is defined by the onset of water loss . in this case , a wet stratosphere
is assumed to exist where water is lost by photodissociation and subsequent hydrogen escape to space .
furthermore , the outer limit of the chz is defined by the first co@xmath3 condensation attained by the onset of formation of co@xmath3 clouds at a temperature of 273 k ; see , e.g. , @xcite and @xcite for further details .
table 3 conveys the results for the hzs for the different types of main - sequence stars of the present study with the different limits referred to as hz@xmath19 . for the outer edge of circumstellar habitability ,
even less stringent limits have been introduced in the meantime ( e.g. , * ? ? ?
* ; * ? ? ?
they are based on the assumption of relatively thick planetary co@xmath3 atmospheres as well as strong backwarming that may further be enhanced by co@xmath3 crystals and clouds .
these limits , which in case of the sun correspond to 2.4 au @xmath20 , conform to the extended habitable zone ( ehz ) , have also been taken into account in our study , although the significance of the ehz has meanwhile been criticized as a result of detailed planetary radiative transfer models @xcite . moreover , in the framework of the present study , we also consider planetary earth - equivalent positions defined as @xmath21 and labelled as @xmath22 ; see table
it is meant as an intriguing reference distance of habitability both regarding single stars and stellar binary systems .
next we introduce the governing equations for investigating the rhzs of binary systems pertaining to both s - type and p - type orbits .
this approach targets the requirement of providing a habitable region for a system planet based on the radiative energy fluxes of the stellar components .
the requirement of planetary orbital stability will be disregarded for now ; it will be revisited in sect .
the importance of orbital stability for allowing circumstellar habitability in stellar binaries will , however , be considered in an appropriate and consistent manner in the main body of the study . for a star of luminosity @xmath23 , given in units of solar luminosity @xmath24
, the distance @xmath25 of the habitability limit @xmath17 as identified for the sun , which may constitute either an inner or outer limit of habitability ( except @xmath26 ) , is given as @xmath27 in case of a multiple star system of order @xmath4 with distances @xmath25 , the limit of habitability related to @xmath17 is given as @xmath28 in eq .
( 2 ) and ( 3 ) , @xmath29 ( see table 1 ) describes the stellar flux in units of the solar constant that is a function of the stellar effective temperature @xmath5 ( e.g. , * ? ? ?
* ; * ? ? ?
specifically , using the formalism represents the normalized stellar flux in units of the solar constant , 1368 w m@xmath30 , given by the stellar spectral energy distribution .
therefore , ordinarily , no @xmath17 dependence for @xmath31 should exist .
however , the formulae by @xcite utilize previous results by @xcite who provided numerical values for limits of habitability for different types of stars considering various limit definitions ( i.e. , @xmath17 values identified for the sun ) . but
@xcite used for the solar effective temperature an unusually low value of 5700 k instead of 5777 k as currently accepted ( e.g. , * ? ? ? * ) . hence , transforming the polynomial fit based on the work by @xcite aimed at considering the correct solar effective temperature renders a weak dependence on @xmath17 for the @xmath31 values .
in contrast , the method by @xcite provides a polynomial fit for @xmath31 without considering the solar temperature revision .
an alternative method has been used by @xcite and subsequent work . in this approach
the polynomial fit by @xcite is corrected via a triangular function based on data for stars of spectral type f0 v , g0 v , and k0
v. as a result the corresponding @xmath31 values do also not depend on @xmath32 .
] by @xcite , we find that @xmath33 with @xmath17 , @xmath34 , and @xmath35 in au , and @xmath36 in k. @xcite also found that for @xmath37 , corresponding to inner limits of habitability , the fitting parameters are given as @xmath38 and @xmath39 , whereas for @xmath40 , corresponding to outer limits of habitability , they are given as @xmath41 and @xmath42 ; note that @xmath43 corresponds to the customary notion of earth - equivalent positions .
appropriate values for @xmath17 are given in table 2 . in the following
we will focus on the case of binary systems , i.e. , @xmath44 . in this case
( 3 ) reads @xmath45 with @xmath46 here @xmath47 denotes the semidistance of binary separation , @xmath48 the distance of a position at the habitability limit contour ( which later on will be referred to as radiative habitable limit " , see below ) , and @xmath49 the associated angle ; see fig . 1 for information on the coordinate set - up for both s - type and p - type orbits .
we will also assume @xmath50 without loss of generality .
with @xmath51 defined as @xmath52 henceforth referred to as _
recast stellar luminosity _
( see table 4 ) , @xmath53 is given as @xmath54 with @xmath55 equation ( 8) constitutes a fourth - order algebraic equation that is known to possess four possible solutions @xcite , although some ( or all ) of them may constitute unphysical solutions , i.e. , @xmath53 having a complex or imaginary value .
the adopted coordinate system constitutes , in essence , a polar coordinate system except that negative values for @xmath48 are permitted ; in this case the position of @xmath48 is found on the opposite side of angle @xmath49 . in principle
, it is possible to consider for eq .
( 8) to only have solutions for @xmath53 given as @xmath56 ; in this case the entire interval for @xmath53 , which is @xmath57 , needs to be examined .
the following types of solutions are identified : for s - type orbits , two solutions exist in the intervals centered at @xmath58 and at @xmath59 ( or one coinciding solution at each tangential point ) ; see fig . 1 .
however , there will be no solution in the typically relatively large intervals containing @xmath60 and @xmath61 .
clearly , the size of any of those intervals critically depends on the system parameters @xmath47 , @xmath17 , @xmath62 , and @xmath63 , as expected . for p - type orbits , on the other hand
, there will be one solution for each value of @xmath49 in the range of @xmath57 .
however , in general , negative values for the solutions of @xmath53 also exist . if taken into account , it will be sufficient to restrict the evaluation of eq .
( 8) to the range @xmath64 . in this case , for s - type orbits , there will be four solutions in the intervals with endpoints @xmath58 and @xmath59 , as well as two solutions ( if @xmath65 ) in a more extended interval containing these points .
also , a pair of solutions will become one coinciding solution at each tangential point .
however , again , there will be no solution in the interval containing @xmath60 . in case of p - type orbits
, there will be two solutions for any value of @xmath49 in the range of @xmath64 .
we will revisit this assessment in conjunction with the algebraic method for attaining the solution @xmath66 ; additionally , detailed mathematical criteria will be given for the existence of rhz for s - type and p - type orbits .
next we will focus on equal - star binary systems .
detailed solutions for general binary systems ( i.e. , systems of stellar components with by default unequal masses , luminosities , and effective temperatures ) pertaining to both s - type and p - type orbits will be given in sect .
3.3 . both subsections will be aimed at deriving rhzs ; see , e.g. , @xcite for general discussions on the role of rhzs for the attainment of habitability in star planet systems .
however , strictly speaking , they will deal with identifying _ radiative habitable limits _ ( rhls )
connected to a distinct value of @xmath17 noting that manifesting a rhz requires that the rhl for @xmath67 to be located _ completely outside _ of the rhl for @xmath68 with @xmath68 and @xmath67 appropriately paired .
a summary about the existence and structure of the rhzs , encompassing the radiative chzs , ghzs , and ehzs , will be given in sect .
3.4 ; this subsection will also convey cases where no rhzs exist due to the behavior of the rhls owing to the choices of @xmath68 and @xmath67 .
now we focus on the special case of equal - star binary systems , i.e. , stars of identical recast luminosities , i.e. , @xmath69 . for theoretical main - sequence stars
this assumption also implies @xmath70 and @xmath71 ; this latter assumption about the stellar masses is relevant for the orbital stability constraint of system planets . with @xmath72 , eq .
( 8) now constitutes a biquadratic equation that can be solved in a straightforward manner .
the other coefficients are given as @xmath73 thus , the solution of eq .
( 8) is given as @xmath74 with @xmath75 with known systems parameters , which are @xmath47 , @xmath17 , and @xmath76 , the function @xmath53 , describing the habitability limits for the binary system associated with inner limit and outer limit values @xmath17 derived for the sun ( see sect . 2 for details )
can be obtained in a straightforward manner .
owing to the system symmetry , the existence of s - type and p - type rhls can be identified by attaining the solutions of eq . ( 11 ) for @xmath77 and @xmath78 .
first we examine the solutions of eq .
( 11 ) for @xmath77 , i.e. , @xmath79 , which are given as @xmath80 this allows us to explore the existence of s - type rhls .
the total number of solutions for @xmath66 ( if existing ) is four as expected , which can be ordered as @xmath81 . due to symmetry
it is found that @xmath82 , which implies that @xmath83 thus , the condition for the existence of s - type rhls is given as @xmath84 next we examine the solutions of eq .
( 11 ) for @xmath78 , i.e. , @xmath85 .
this allows us to explore the existence of p - type rhls ; the latter implies two solutions of eq . (
11 ) regardless of the value for @xmath49 . if the positive root of @xmath86 ( see eq .
12 ) is considered , the solution is given as @xmath87 thus , the condition for the existence of p - type rhls is given as @xmath88 therefore , eqs .
( 15 ) and ( 17 ) allow to identify the conditions for s - type and p - type rhls , respectively , for equal - star binary systems , which depend on the systems parameters @xmath47 , @xmath17 , and @xmath76 ; note that the equal signs in these equations carry little relevance . comparing eqs .
( 15 ) and ( 17 ) also implies that the joint existence of s - type and p - type habitability in equal - star binary systems in circular orbits is not possible , irrespectively of the system parameters and the planetary orbital stability requirement ( see sect . 4 ) , noting that the latter imposes an additional constraint on habitability even when the rhz - related conditions are met .
figure 2 depicts the borders of the s - type and p - type radiative habitable limits , i.e. , rhls , for different values of @xmath17 in regard to @xmath47 and @xmath76 .
we now focus on obtaining solutions for our key equation , eq .
( 8) , pertaining to s - type and p - type rhls for general binary systems , i.e. , @xmath65 . following , e.g. , @xcite and @xcite , the set of solutions for a fourth - order polynomial reads @xmath89 with @xmath90 with @xmath91 as a solution of the resolvent cubic equation @xmath92 with @xmath93 , @xmath94 , and @xmath95 given by eqs .
( 9a ) to ( 9c ) ; here the term of @xmath96 in eqs .
( 19b ) and ( 19c ) corresponds to the case of non - equal star binaries assumed in the following .
note that for equal - star binaries a more straightforward method of solution is available ( see sect .
the solutions @xmath66 given through eqs .
( 18a ) to ( 18d ) , if existing , are ordered as @xmath81 for @xmath58 ; this order is also maintained for any other value of @xmath49 as identified in all model simulations pursued .
although eq .
( 20 ) has three possible solutions , there is only one appropriate choice for @xmath97 , named @xmath91 , because it is necessary to avoid that all @xmath66 obtained through eq .
( 8) are of imaginary or conjugate complex value in cases where s - type or p - type rhls exist .
the acceptable solution for @xmath97 is given as @xmath98 with the substitutions @xmath99{r + \sqrt{d_3 } } \\
t & = & \sqrt[3]{r - \sqrt{d_3 } } \\ d_3 & = & q^3 + r^2\end{aligned}\ ] ] and with @xmath100 and @xmath101 given as @xmath102 while noting that @xmath103 these sets of equation can be solved and appropriate values for @xmath53 can be obtained .
the results will depend on the system parameters @xmath47 , @xmath17 , @xmath62 , and @xmath63 , as expected .
next we describe the solutions for s - type and p - type rhls in more detail .
it is important to recognize that a priori choices about the existence of s - type and p - type rhls are neither necessary nor possible as the existence of any of those rhls is determined by the fulfillment of well - defined mathematical conditions ; they will also be given in the following .
an analysis of the possible solutions for eqs .
( 18a ) to ( 18d ) shows that for s - type rhls valid solutions are obtained based on @xmath104{r^2 + k^2 } \
\big ( \cos \xi + i \sin \xi \big ) \\ t & = & \sqrt[6]{r^2 + k^2 } \
\big ( \cos \xi - i \sin \xi \big)\end{aligned}\ ] ] with @xmath105 with @xmath101 given by eq .
( 23b ) . therefore , the solution of the resolvent cubic equation , eq .
( 20 ) , is given as @xmath106{r^2 + k^2 } \cos \xi\ ] ] for the values of @xmath66 it is found that for the interval centered at @xmath58 , @xmath107 and @xmath108 exhibit negative values , whereas @xmath109 and @xmath110 exhibit positive values with @xmath81 . thus , @xmath107 and @xmath108 describe the rhl regarding star s1 , whereas @xmath109 and @xmath110 describe the rhl regarding star s2 ( see fig . 1 ) .
conversely , for the interval centered at @xmath59 , @xmath107 and @xmath108 again exhibit negative values and @xmath109 and @xmath110 exhibit positive values . in this case ,
@xmath107 and @xmath108 describe the rhl for star s2 , whereas @xmath109 and @xmath110 describe the rhl for star s1 .
no solutions are obtained in the vicinity of @xmath60 and @xmath61 , as expected .
thus , for each angle @xmath49 in the range of @xmath57 the appropriate number of solutions is attained to describe s - type rhls . however , due to symmetry , solutions are only needed for @xmath64 .
the existence of s - type rhls requires that @xmath111 because otherwise the two distinct s - type rhl contours about the two binary components would not be separated , which corresponds to the condition @xmath112 equation ( 28 ) can be rewritten to provide an expression based on the system parameters @xmath93 , @xmath94 , and @xmath95 defined through eqs .
( 9a ) to ( 9c ) .
it is found that @xmath113 with @xmath114 .
the expression for @xmath91 is highly complicated ; however , it can be obtained based on eqs .
( 21 ) to ( 24c ) by using , e.g. , mathematica@xmath115 in a straightforward manner . in conclusion , for s - type rhls to exist for the system parameters @xmath47 , @xmath17 , @xmath62 , and @xmath63 ,
it is necessary that the relations ( 28 ) and ( 29 ) , which are equivalent , must be fulfilled for any angle of @xmath49 , though the evaluation can be limited to @xmath77 .
furthermore , through analytical transformations it can be shown that the condition depicted as eqs .
( 28 ) and ( 29 ) requires @xmath116 in the limiting case of equal - star binary systems , attained as @xmath117 , eq .
( 30 ) can be simplified as @xmath118 this relationship is fulfilled in a trivial manner .
an analysis of the possible solutions for eqs .
( 18a ) to ( 18d ) also shows that for p - type rhls valid solutions require @xmath119 see eq .
the detailed evaluation of this condition requires the evaluation of various sets of equations denoted as eqs .
( 9a ) to ( 9c ) , ( 22a ) to ( 22c ) , and ( 23a ) to ( 23b ) ; see sect .
3.1 and 3.3.1 . in terms of the solutions for p - type rhls it is found that for @xmath120 and @xmath121 , @xmath107 exhibit negative values and @xmath108 exhibit positive values , whereas @xmath109 and @xmath110 are undefined
; they are also not needed for outlining p - type rhls . moreover , for the range of @xmath122 , @xmath109 exhibit negative values and @xmath110 exhibit positive values , noting that @xmath107 and @xmath108 remain undefined . for @xmath123 and @xmath124 ,
removable singularities are identified , which can easily be fixed through interpolation taking values of @xmath48 for neighboring angles of @xmath49 . in summary , for each angle @xmath49 in the range of @xmath57 two values of @xmath66 ( i.e. , one positive and one negative value ) are identified allowing to determine p - type rhls . however , due to symmetry , solutions are only needed for @xmath64 .
moreover , through analytical transformations it can be shown that the condition depicted as eq .
( 32 ) can be rewritten as @xmath125 with @xmath93 , @xmath94 , and @xmath95 defined through eqs .
( 9a ) to ( 9c ) with the left hand side of eq .
( 33 ) representing @xmath126 . in conclusion , for p - type rhls to exist for the system parameters @xmath47 , @xmath17 , @xmath127 , and @xmath128 , it is necessary that the relations ( 32 ) and ( 33 ) , which are equivalent , must be fulfilled for any angle of @xmath49 , though the evaluation can be limited to @xmath78 . in the limiting case of equal - star binary systems , attained as @xmath117 , eq .
( 33 ) can be simplified as @xmath118 this relationship , already given as eq .
( 31 ) , is fulfilled in a trivial manner .
the identification of the rhzs in binary systems requires the calculation of limits of habitable zones , i.e. , rhls , as pointed out in sect .
the rhzs need to be established for values of @xmath17 with @xmath18 = 1 , 2 , 4 , 5 , and 6 ( see sect . 2 and table 2 ) , which are informed by model - dependent physical limits of habitability for the solar environment ( e.g. , * ? ? ?
as part of the process , the parameters of @xmath17 need to be appropriately paired in terms of the inner and outer limits of habitability . for the chz the parameters @xmath129 need to be paired as @xmath130 , whereas for the ghz , they need to paired as @xmath131 .
for the ehz , the parameters of @xmath17 need to paired as @xmath132 , considering that both the chz and the ghz shall be viewed as subdomains of the ehz . for s - type and p - type orbits , the radiative zones of habitability @xmath133 , which constitutes a circular region ( annulus ) around each star s1 and s2 ( s - type ) or both stars ( p - type ) can be determined as @xmath134 and @xmath135 respectively
; see fig . 1 for coordinate information .
here @xmath136 and @xmath137 describe the areas bordered by the rhls defined by @xmath138 and @xmath139 .
the calculation of the extrema is applied to the angles @xmath140 and @xmath49 for the intervals @xmath141 and @xmath142 , respectively ; note that we assumed @xmath50 without loss of generality . in the s - type case
the calculation of the extrema pertaining to the rhz values is based on the angular coordinate @xmath140 instead of @xmath49 ; however , the angular coordinate @xmath49 is still needed for the calculation of @xmath66 as part of the overall approach toward identifying s - type and p - type habitability .
figure 3 depicts examples of rhls and rhzs for different types of systems . in the s - type case
the rhls are bended toward the center of the system , whereas in the p - type case they are of notable elliptical shape .
the rhzs always constitute circular annuli obtained through inspecting the appropriate minima and maxima of the rhls .
the examples as depicted include s - type and p - type systems with separation distances @xmath143 of 0.5 au and 5.0 au , respectively .
cases of both equal - star and non - equal star binaries are selected .
the focus of this figure is the identification of the appropriate circular region ( i.e. , annulus ) for each case .
the figure also indicates the portions within the @xmath144 and @xmath145 domains that are not part of the rhz@xmath146 annuli .
next we determine the values for the extrema pertaining to @xmath133 following eqs .
( 35 ) and ( 36 ) based on the solutions of eq .
( 8) given as eqs .
( 18a ) to ( 18d ) . in cases
where four solutions @xmath66 exist , it is found that they are ordered as @xmath81 with @xmath107 and @xmath108 constituting negative values , and @xmath109 and @xmath110 constituting positive values . if the negative solutions for @xmath66 are permitted , it is sufficient for s - type orbits , both for star s1 and s2 , to only consider solutions for @xmath58 . for p - type orbits ,
a more detailed assessment is required ( see below ) . for s - type orbits , regarding star s1 ,
the extrema are obtained as @xmath147 and for star s2 , they are obtained as @xmath148 the size of each annulus @xmath149 for pairs @xmath129 is given as @xmath150 .
@xmath151 also constitutes a generalization of hz@xmath19 with @xmath152 previously defined for single stars ( see sect . 2 and table 3 ) .
likewise , @xmath153 constitutes a generalization of hz@xmath19 with @xmath154 4 , 5 , and 6 . for p - type orbits ,
the extrema are given as follows : @xmath155 we note that for the angle @xmath156 for @xmath153 no straightforward expression , an analytic expression for @xmath156 is deemed possible .
however , it will be highly complicated and thus a numerical solution may be preferred . ]
exists ; it is located in the interval @xmath157 .
it can be found numerically as it is given by the angle where the minimum of @xmath158 occurs . in the special case of @xmath159 with @xmath62 and @xmath63 denoting the stellar primary and secondary , respectively , it is found that @xmath160 , whereas for @xmath161 it is found that @xmath162 ( see sect . 3.2 ) .
there is also another complication in the identification of rhz p - type orbits .
generally , it is required for the rhl of @xmath67 to be located completely outside of the rhl of @xmath68 , i.e. , @xmath163 this condition is however violated in some models , especially for relatively large values of @xmath47 as well as relatively small ratios of @xmath164 . in this case
the rhz for @xmath129 is nullified , a behavior that may occur for the pairings @xmath130 , @xmath131 , and @xmath132 , corresponding to the chz , ghz , and ehz , respectively . in this regard
the existence of the chz is in most jeopardy as @xmath165 constitutes the smallest bracket among the various kinds of hzs ( see table 2 ) .
also note that if the ghz is nullified , the chz will be nullified as well considering that the chz ( if existing ) is entirely located within the ghz .
likewise , if the ehz is nullified the existence of both the chz and the ghz will be nullified .
detailed examples will be given in the application segment of this paper ; see sect .
5.2 for details .
however , this type of phenomenon does not occur for rhzs pertaining to s - type orbits . for equal - star binary systems , with the property of @xmath69 , expressions for rhz@xmath166 and rhz@xmath167 for s - type and p - type orbits can be obtained based on eqs .
( 13 ) and ( 16 ) . for s - type orbits
we find @xmath168 it is also intriguing to explore the limits @xmath169 and @xmath170 .
if these limits are met , it is found that @xmath171 moreover , in the limit of @xmath172 , the expressions for single star hzs regarding @xmath151 and @xmath153 are recovered , as expected , which are given as @xmath173 and @xmath174 , respectively .
they are in agreement with the expressions previously obtained by @xcite , @xcite , @xcite , and others . results for p - type orbits can be obtained considering @xmath175 in this case we find @xmath176 based on the system parameters @xmath47 , @xmath68 , @xmath67 , @xmath177 , and @xmath178 . additionally , the requirement to avoid that the rhl for @xmath67 to be partially or completely located inside of the rhl for @xmath68 , see eq .
( 40 ) , entails @xmath179 which allows to set constraints on the separation distance @xmath143 of the binary system noting that the values of @xmath68 , @xmath67 , @xmath177 , and @xmath178 are subject to distinct restrictions , particularly in case of main - sequence stars ( see tables 1 and 2 ) .
depictions of the condition ( 45 ) for equal - star binaries for the pairings @xmath130 , @xmath131 , and @xmath132 , corresponding to the chz , ghz , and ehz , respectively , are given in fig .
a similar expression is expected to hold for nonequal - star binaries , albeit it will be highly complicated .
hence , for those systems a numerical assessment of @xmath151 and @xmath153 ( see eqs .
39a and 39b ) may be preferred to accommodate condition ( 40 ) ; see sect .
5.2.2 for additional information and data .
a primary constraint on planetary habitability is that planets are required to exist in the hz for a sufficient amount of time allowing basic forms of life to emerge and develop . in order to adhere to this criterion
, planetary orbital stability is required .
there is a significant body of literature devoted to this topic , including studies of binary and multi - planetary systems , which often also consider aspects of stellar evolution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
early studies of planetary orbital stability pertaining to planets in both s - type and p - type orbits demonstrated that planets can exist in systems of binary stars for 3000 binary periods @xcite .
although these investigations considered relatively short integration times , @xcite determined upper and lower bounds of planetary orbital stability considering the orbital elements , semimajor axis and eccentricity , of the proposed binary stars . since this pioneering work
, many additional studies have been performed .
the foremost investigation extended the original study by a factor of 10 in integration times and an extended range of orbital elements @xcite .
in addition , the nature of the bounding formula was derived and discussed using a more statistical framework .
@xcite developed fitting formulae for both s - type and p - type planets in binary systems given as @xmath180 and @xmath181 respectively .
these equations give the critical semimajor axis @xmath182 in units of the semimajor axis @xmath47 in case of s - type and p - type orbits .
for an s - type orbit , the ratio @xmath183 , see eq .
( 46 ) , conveys the _ upper limit _ of planetary orbital stability , whereas for a p - type orbit , the ratio @xmath183 , see eq . ( 47 ) , conveys the _ lower limit _ of planetary orbital stability .
moreover , @xmath184 denotes the stellar mass ratio given as @xmath185 , where @xmath186 and @xmath187 constitute the two masses of the binary components with @xmath188
. equations ( 46 ) and ( 47 ) also contain the parameter functions @xmath189 and @xmath190 , which depend on the aforementioned mass ratio @xmath184 and the eccentricity of the stellar binary , @xmath191 . considering
that this paper is solely aimed at stellar binaries in circular orbits ( i.e. , @xmath192 ) , it is found that @xmath193 .
planetary orbital stability has been investigated by many authors using chaos indicators , such as the maximal lyapunov exponents ( mle ) , fast lyapunov indicator ( fli ) , and the mean exponential growth factor of nearby orbits ( megno ) , to name those commonly used ; see , e.g. , @xcite for details , recent applications , and references .
these methods have also been used to characterize the transition from stable to unstable orbits within the framework of the circular and elliptical 3-body problems ; see , e.g. , @xcite , @xcite , and @xcite for details .
previously , @xcite studied the stability of both s - type and p - type orbits in stellar binary systems , and deduced orbital stability limits for planets .
these limits were found to depend on the mass ratio between the stellar components and the distance ratio between planetary and binary semimajor axes .
this topic was revisited by @xcite , who used the concept of jacobi s integral and jacobi s constant to deduce stringent criteria for the stability of planetary orbits in binary systems for the special case of the coplanar circular restricted three - body problem .
recently the planetary orbital stability was studied through the perspective of a chaos indicator , the mle by , e.g. , @xcite . from the use of a chaos indicator a cutoff value for the maximum lyapunov exponent
was determined as an additional stability criterion for s - type planets in the circular restricted 3-body problem .
next we investigate s - type habitability for selected binary systems , including systems of equal and non - equal masses ( see table 5 ) .
our main intent is to demonstrate the functionality of the method - as - proposed and @xmath194 , and a saturnian planet in a p - type orbit ; see @xcite for a detailed study of the system s habitability . ] ; an extensive parameter study will be given in sect .
6 . figure 5 allows comparative insight into s - type habitability for selected binary systems , i.e. , systems with masses of @xmath195 and @xmath196 , @xmath197 ; the binary separation distances are chosen as 10 au and 20 au . for single stars of @xmath198 , the radiative chz extends from 1.049 to 1.498 au , and the radiative ghz extends from 0.927 to 1.831 au ; these values are slightly higher than those for g2 v stars given in table 3 owing to a minuscule difference in mass ( i.e. , 0.99 versus 1.0 @xmath199 ) . for an equal - mass binary system of 1.0 @xmath199 with a separation distance @xmath143 of 10 au ,
the radiative chz and ghz extend from 1.056 to 1.511 au , and from 0.932 to 1.853 au , respectively , for each component .
furthermore , the outer limit of the radiative ehz is altered from 2.64 to 2.70 au .
however , there is now an upper orbital stability limit of 1.37 au imposed on each star .
consequently , significant portions of the radiative chz and ghz are unavailable as circumstellar habitable regions .
if the second star is placed at a distance of 20 au , the alteration of the radiative chz and ghz relative to single stars is very minor . specifically , for binary separations of 10 au and 20 au ,
the sizes of the radiative ghz increase by 1.9% and 0.6% relative to the case of single stars .
moreover , for the system with a separation distance of 20 au , the imposed orbital stability limit is found at 2.74 au ; consequently , the full extents of the radiative chz , ghz , and ehz are now available for planetary habitability .
figure 5 also shows results for the pairs @xmath200 @xmath199 and @xmath201 @xmath199 . in case of a single 1.5 @xmath199 mass star ,
the radiative chz and ghz extend from 1.88 to 2.49 au , and from 1.65 to 3.11 au , respectively , whereas the radiative ehz extends up to 4.61 au . in this type of system ,
a secondary star of 1.0 @xmath199 placed at a separation distance of 10 au again modifies the extents of the radiative chz and ghz , which now extend from 1.89 to 2.49 au and from 1.66 to 3.14 au , respectively ; however , the planetary orbital stability limit now occrs at 1.56 au .
therefore , the entire domains of the radiative chz , ghz , and ehz of the primary are unavailable as circumstellar habitable regions .
if the secondary star is placed at a separation distance of 20 au , the radiative chz , ghz , and ehz of the primary star are again similar to those of a 1.5 @xmath199 mass star .
however , the orbital stability limit is now found at a distance of 3.12 au from the primary ; therefore , the entire supplement of the radiative ehz , given by the bracket @xmath202 is now considered habitable . in summary , for potentially habitable s - type binaries , owing to the implied requirement of the relatively large separations of the stellar components , the effect of
the stellar secondary on the extents of the rhzs is often minor , i.e. , about a few percent or less , with the biggest impact occurring in f - type systems .
for most systems , the secondary s main influence on circumstellar habitability thus consists in limiting planetary orbital stability rather than offering significant augmentations of the rhzs , a feature most pronounced in close binaries .
various sets of models have been pursued to examine p - type habitability ( see fig .
6 ) . as examples we considered systems with masses of @xmath195 and @xmath196 and @xmath203
; additionally , we also focused on models of @xmath204 and @xmath205 ( see tables 5 to 8 for details ) .
the separation distances @xmath143 were chosen as 0.5 , 1.0 , and 2.0 au , respectively .
our approach consists again of two steps .
first , we explore the existence and extent of the radiative chzs , ghzs , and ehzs .
subsequently , we consider the additional constraint of planetary orbital stability , which in the case of p - type orbits constitutes a lower limit ( see sect . 4 ) .
our results can be summarized as follows . for systems with masses of @xmath195 ,
the following behavior is found .
for separation distances @xmath143 of 0.5 au , the inner limit ( i.e. , rhl ; see sect .
3.4 ) of the radiative chz varies between 1.46 and 1.54 au as function of polar angle @xmath49 with 1.54 au to be considered as acceptable inner limit ; see eq .
furthermore , the outer limit of the radiative chz varies between 2.10 and 2.16 au with 2.10 au as acceptable outer limit ; see eq .
( 39b ) . in consideration of the orbital stability limit at 0.92 au ( see eq .
45 ) , constituting an inner limit of orbital stability , the entire extent of the radiative chz is available as a circumbinary habitable region
. the acceptable inner limit of the radiative ghz is given as 1.38 au , whereas the acceptable outer limit occurs at 2.58 au ; hence , the entire radiative ghz is again identified as habitable . for separation distances of 1.0 au ,
the orbital stability limit is given at 1.83 au , which falls inside the domain of the radiative chz ranging from 1.69 to 2.06 au ; therefore , only about half of the radiative chz is available for circumbinary habitability , whereas the other half is not . since the radiative chz is fully embedded into the radiative ghz , only a fraction of the radiative ghz offers circumbinary habitability .
however , the full extent of the supplementary radiative ehz , given by the bracket @xmath202 , with an acceptable outer limit of 3.83 au , offers habitability .
we also considered models with binary separations of 2.0 au . in this case , the orbital stability limit is found at 3.66 au .
therefore , both the radiative chz and ghz are unavailable for providing habitability ; the latter has an outer limit that varies between 2.39 and 3.05 au with 2.39 au as acceptable limit .
the outer limit of the radiative ehz varies between 3.60 and 4.09 au with 3.60 au to be ruled acceptable as the conservatively selected ( i.e. , inner ) limit of the radiative ehz .
hence , the entire radiative ehz is also not considered available for providing circumbinary habitability .
most significantly , we also pursued case studies for systems of unequal distributions of mass , and by implication of unequal distributions of luminosity as , for example , the system @xmath196 and @xmath203 .
according to the mass luminosity relationship for main - sequence stars , it is found that a 1.5 @xmath199 star possesses a luminosity @xmath7 about 3.5 times higher than a 1.0 @xmath199 star ; a similar factor of difference exists for the recast stellar luminosity @xmath206 ( see tables 4 and 5 ) .
thus , the combined luminosity of the @xmath207 system is considerably higher than the combined luminosity of the @xmath208 system , as expected . on the other hand ,
following the work by @xcite and @xcite , an unequal distribution of stellar mass , i.e. , a smaller value of @xmath184 ( see sect . 4 ) , entails a smaller orbital stability limit .
since it constitutes a lower limit , i.e. , positioned more closely to the stellar system , it offers larger
windows of opportunity " for planets in the rhzs ( if existing ) to be orbitally stable .
results for separation distances @xmath143 of 0.5 , 1.0 , and 2.0 au are given in fig .
6 . for a binary separation of 0.5
au , it is found that both the radiative chz and ghz exist , and habitability in these domains is fully permitted according to the planetary orbital stability constraint , although the width of the chz is relatively small .
the chz extents from 2.14 to 2.28 au , whereas the ghz extents from 1.91 to 2.90 au ; the orbital stability limit is given at 0.66 au . in this type of system , there are extreme variations for the inner and outer limits of both the radiative chz and ghz . for example
, the inner rhl for the chz varies between 1.65 and 2.14 au , whereas its outer rhl varies between 2.28 and 2.76 au as function of polar angle @xmath49 .
there is also a considerably large domain of the supplementary portion of the radiative ehz , which has an outer limit that varies between 4.40 and 4.89 au .
detailed depictions of the variations of the inner and outer limits of the rhzs for the various systems are given in fig .
this figure indicates relatively small bars of variations for equal - mass systems such as @xmath195 with small separation distances as , e.g. , @xmath209 au .
however , large bars of variations are obtained for non - equal mass binaries or for equal - mass binaries with large separation distances as , e.g. , @xmath210 au . in systems with a binary separation of 1.0 au , the radiative chz is nullified ; note that the orbital stability limit in this system is given at 1.32 au .
the reason for the disallowance of the chz is that the rhl for @xmath211 , which is at 2.38 au , is located inside of the rhl for @xmath212 , given as 2.05 au . the same criterion ( see eq .
40 ) also leads to a relatively small width of the radiative ghz , which extends between 2.16 and 2.66 au . at a binary separation of 2.0 au
, the situation is even more drastic as both the radiative chz and ghz are disallowed .
the only type of circumbinary habitable region remaining is that provided by the relatively large supplementary portion of the radiative ehz given by the bracket @xmath202 . in this zone
, habitable planets are expected to be possible as their existence would be consistent with the planetary orbital stability constraint .
next we explore the existence of p - type rhzs , both for equal - mass and non - equal mass binaries , in a more systematic manner through the means of _ numerical experiments_. specifically , we pursue sets of model calculations with the binary separation distance @xmath143 considered as an independent variable ; see table 9 for results .
the stellar masses are altered between @xmath213 and @xmath214 in increments of @xmath215 ( see table 5 ) .
results are given for the pairings @xmath130 ( chz ) , @xmath131 ( ghz ) , and @xmath132 ( ehz ) .
note that for equal - mass binaries , it is sufficient to solve eq .
( 45 ) , whereas for general binary systems a more thorough assessment is needed to satisfy relation ( 40 ) .
this approach allows us to explore the maximum binary separation distances , which are upper limits for permitting rhzs for each case ( i.e. , combination of binary masses and choice of chz , ghz , or ehz ) .
generally , it is found that for any binary system , the greatest permissible binary separation distance is attained for the ehz , and furthermore that value - as - attained is greater for the ghz than for the chz ; these findings are as expected .
for example , for the system @xmath195 , the expiration distance for the radiative ehz is given as 4.25 au , whereas for the radiative ghz and chz , the distances are given as 2.57 and 1.64 au , respectively ( see table 9 ) .
as another example , the system @xmath204 and @xmath205 , the expiration distances for the radiative ehz , ghz , and chz are given as 3.80 , 1.93 , and 0.96 au , respectively .
it is also intriguing to compare results for stellar pairs as , e.g. , @xmath195 to stellar pairs such as @xmath196 and @xmath203 .
for the system of @xmath195 , it is found that the radiative chz and ghz are nullified as defined by the limit of validity of eq .
( 45 ) at binary separation distances of 1.64 and 2.57 au , respectively ( see table 9 ) . at those binary separations , the distances of the vanishing rhz - chz and rhz - ghz ( as measured from the geometrical center of the system
; see fig . 1 )
are given as 1.95 and 2.25 au , respectively . in comparison ,
the limits of planetary orbital stability ( to be interpreted as lower limits ) are identified as 3.00 and 4.71 au , respectively .
thus , we conclude that for equal - mass binary systems such as @xmath195 , habitability for widely spaced binaries is lost due to the lack of orbital stability already at binary separations where the circumbinary chz - rhzs and ghz - rhzs are still in place .
the same type of study has been pursued for systems with highly unequal intrabinary distributions of masses and , by implication , stellar luminosities as , e.g. , @xmath196 and @xmath203 . in this case
it is found that the radiative chz and ghz vanish at binary separation distances @xmath143 of 0.65 and 1.55 au , respectively .
furthermore , the distances of the vanishing radiative chz and ghz ( as measured from the system center , see fig .
1 ) are given as 2.21 and 2.43 au , respectively ( see table 9 ) .
the respective limits of planetary orbital stability are identified as 0.85 and 2.04 au .
thus , for this type of system it is found that habitability is lost due to the vanishing rhzs , even though circumstellar habitability would still be permitted according to the planetary orbital stability criterion .
the fact that circumbinary habitability is lost already for systems of relatively small binary separations is a consequence of the extreme radiative imbalance caused by the highly unequal distribution of stellar luminosities , which determine the circumbinary rhls .
radiative imbalance within binary systems may cause the rhl for @xmath67 to be partially or completely located inside of the rhl for @xmath68 ; see sect .
in fact , when the pairs @xmath195 and @xmath196 and @xmath203 are compared to one another , it is found that although the unequal - mass binary system has almost twice the combined stellar luminosity of the equal - mass binary system ( i.e. , 3.85 versus 2.0 @xmath24 ) , it still possesses much narrower chz , ghz , and ehz rhzs .
in fact , it is found that the condition expressed as eq .
( 40 ) is most readily met in cases of equal - mass binary systems of relatively small separation distances and mostly violated in systems of relatively large separation distances and/or unequal distributions of masses and , by implication , luminosities .
various examples have been depicted in fig . 7 ; see discussion in sect .
5.2.1 . in summary ,
although an unequal distribution of stellar masses within binary systems is identified as advantageous for facilitating planetary orbital stability , in consideration of that lower stability limits for p - type orbits occur for smaller mass ratios @xmath184 ( see sect . 4 ) , the situation for the existence of the rhzs is much less ideal , even for systems where the stellar primary is highly luminous owing to the behavior of the rhls . in this regard ,
the radiative chz is in most jeopardy as @xmath165 constitutes the smallest bracket among the various kinds of hzs ( see table 2 ) .
more fortunate scenarios are expected to occur for the radiative ghz and ehz , with the brackets given as @xmath216 and @xmath217 , respectively ; they are characterized by considerably larger widths , especially in case of equal - mass systems of stars with relatively high luminosities .
another aspect of this study is to provide an appropriate classification of habitability applicable to general binary systems .
previously , @xcite introduced the terminology of s - type and p - type orbits for system planets , which is now widely used by the orbital stability , planetary , and the astrobiology science communities .
evidently , besides the assessment of orbital stability behaviors , these terms are also appropriate for classifying binary system rhzs , if existing . however ,
following previous investigations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , as well as the results of the present work , the spatial domain of s - type and p - type habitability depicted by the rhzs is often adversely affected , and in some cases even nullified , by the requirement that system planets must be orbitally stable .
thus , if the available extent of the s - type and p - type rhzs for the manifestation of habitability is truncated owing to the additional constraint of planetary orbital stability , these zones shall be referred to as st - type and pt - type , respectively , in the following .
detailed results are given in table 10 , which provides an extensive summary of p , pt , st , and s - type habitability for both equal - mass and non - equal mass binary systems .
the stellar masses are varied between @xmath213 and @xmath214 in increments of @xmath215 amounting to a total of 15 combinations .
table 10 features the results for the pairings @xmath130 ( chz ) , @xmath131 ( ghz ) , and @xmath132 ( ehz ) . in principle
, it is found that with the secondary taken as fixed the higher the mass and , by implication , the luminosity of the stellar primary , the larger values are obtained for p , pt , st , and s - type habitability . additionally , larger values for the limits of p ,
pt , st , and s - type habitability are obtained regarding the ghz relative to the chz , as expected .
the largest values are obtained for pt and s - type habitability for the ehz ; in this regard , there is no change for p and st - type habitability relative to the ghz since both types of hzs are based on the same inner bracket value of @xmath218 ( see above ) .
the results of table 10 are in line with the previously discussed findings about highly unequal intrabinary distributions of masses and , by implication , stellar luminosities as , e.g. , @xmath204 and @xmath205 or @xmath196 and @xmath203 compared to the case of @xmath195 . for systems of highly unequal mass distributions , the domains of p - type and pt - type habitability are typically relatively small as the rhl for @xmath67 crosses the rhl for @xmath68 in relatively close proximity to the primary , such allowing only small distance ranges to exhibit p / pt - type habitability .
it is also found that in seven cases for the chz , as well as two cases for the ghz , the rhzs expire prior to the truncation of habitability due to the planetary orbital stability requirement . in those cases ,
only p - type habitability exists ; no pt - type habitability is found as the orbital stability constraint bears no relevance .
figure 8 and 9 show various combinations of equal - mass and nonequal - mass binary systems ; they all show numerous similarities , though the spatial scales are noticeably different as they are defined through the stellar luminosities . if equal - mass binary systems are considered , taking @xmath195 and @xmath219 as examples , the extent of both p and pt - type habitability increase with increasing stellar mass or luminosity . the distances for p and pt habitability
are found to almost coincide indicating that the orbital stability constraint affects the inner and outer limit of p - type habitability in about the same manner .
furthermore , the inner and outer limits of both s and st habitability are shifted to larger distances from each stellar component for stars of higher luminosity , as expected . moreover , for stars of higher luminosity , there is a larger spatial domain where s - type habitability is truncated due to the additional constraint of planetary orbital stability . for equal - mass systems of @xmath198 ,
st and s - type habitability is identified at distances of 6.85 and 13.46 au , whereas for @xmath213 , st and s - type habitability is identified at 1.38 and 3.00 au .
figure 9 depicts two selected cases of nonequal - mass binary systems . in both cases
the stellar primary is chosen as 1.0 @xmath199 , whereas the stellar secondary is chosen as 0.75 @xmath199 and 0.5 @xmath199 , respectively ; the corresponding stellar luminosities of the secondaries are 0.357 and 0.045 @xmath24 , respectively ( see table 5 ) .
a reduced luminosity of the secondary binary component adversely affects the extent of the rhz , as expected .
interestingly , a reduction of mass for the stellar secondary has a nontrivial impact on the orbital stability domains , which considerably depend on the mass ratio @xmath184 ( see eqs . 46 and 47 ) .
if @xmath184 is reduced , the permissible stability domain for p - type orbits is increased , whereas the permissible stability domain for s - type orbits is decreased .
thus , the assessment of s , p , st , and pt habitability for nonequal - mass binaries requires a detailed computational analysis . in the example of fig . 9 , in regard to @xmath205 , s , p , st , and pt habitability occurs at distances of 0.82 , 1.16 , 6.18 , and 12.19 au , respectively , and for @xmath203 , s , p , st , and pt habitability occurs at distances of 0.94 , 1.01 , 5.50 , and 10.86 au , respectively .
the respective differences are notable , but not drastic ; more pronounced differences occur for systems of more luminous stars ( see table 10 ) .
note that both figs .
8 and 9 refer to habitability assessments pertaining to the ghz ( see table 2 ) .
they are given to exemplarily showcase the structure , extent and location of the s - type and p - type rhzs as well as the relevance of the orbital stability limits for both s - type and p - type habitability , thus allowing us to uniquely identify the spatial domains of s , p , st , and pt - type habitability for each system .
computationally , the occurrence of p , pt , st , and s - type habitability can be identified as follows . for sufficiently small binary separations
@xmath143 , it is found that both the inner and outer limit of the p - type rhz are located beyond of the p - type orbital stability limit ( see eq .
47 ) , which constitutes a lower limit of planetary orbital stability ( see sect . 4 ) . if the distance of binary separation is increased , the inner and outer limits of the p - type rhz decrease , whereas the p - type orbital stability limit increases .
thus , starting at a certain value of @xmath143 , only a fraction of the width of the p - type rhz will be available for providing habitability ; in this case , pt - type habitability is attained .
if the binary separation is further increased , the entire width of the p - type rhz will be unavailable for providing habitability because habitability would be incompatible with the orbital stability constraint .
eventually , the p - type rhz expires ; see also information provided in table 9 . for mid - sized values of the binary separation distance ,
the s - type rhz is encountered to exist , but it is unable to provide habitability because of the s - type orbital stability limit ( see eq .
46 ) , which constitutes an upper limit of planetary orbital stability ( see sect . 4 ) .
the s - type rhzs continue to exist further out ; note that the inner and outer limits of the s - type rhzs essential continue to run parallel as function of the binary separation distance for most systems .
if the binary separation distance @xmath143 is further increased , again as part of our numerical experiment , the s - type orbital stability will increase and will cross the inner limit of the s - type rhz ; in this case , st - type habitability is encountered as some , but not all of the width of the s - type rhz is available for providing habitability . eventually , for sufficiently large binary separations , the s - type orbital stability limit also crosses the outer limit of the s - type rhz . in this case
, the full width of the s - type rhz is available for facilitating habitability , consistent with the definition of s - type habitability .
another application is displayed in fig .
it shows for a given spectral type of equal - star binaries the stellar separation distances @xmath143 for which chzs , ghzs , and ehzs are able to exist .
the chzs , ghzs , and ehzs can be either s or st - type , on one hand , or p or pt - type , on the other hand , to qualify for depiction .
the results are given as function of stellar spectral type , for stars between spectral type f0 to m0 .
the figure shows that p / pt - type habitable regions are able to exist for a relatively large range of separation distances in case of relatively luminous stars ( i.e. , spectral type f ) , but only for a relatively small range of separation distances for lesser luminous stars ( i.e. , spectral types k and m ) .
regarding s / st - type habitable regions the situation is reversed .
figure 10 also indicates a notable domain of binary separations where no habitable regions are found owing to the lack of rhzs , the lack of planetary orbital stability , or both . moreover , no domain of binary separation distances is identified where s / st - type and p / pt - type habitable regions overlap .
in this study we present a new method about a comprehensive assessment of s - type and p - type habitability in stellar binary systems .
p - type orbits occur when the planet orbits both binary components , whereas in case of s - type orbits the planet orbits only one of the binary components with the second component considered a perturbator .
an important characteristic of the new method is that it combines the orbital stability constraint for a system planet with the necessity that a habitable region given by the stellar radiative energy fluxes ( radiative habitable zone " ) must exist .
the requirement to combine these two properties has also been recognized in previous studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
another element of the present study is to introduce a habitability classification regarding stellar binary systems , consisting of habitability types s , p , st , and pt . this type of classification also considers whether or not s - type and p - type radiative habitable zones are reduced in size due to the additional constraint of planetary orbital stability . in summary ,
five different cases were identified , which are : s - type and p - type habitability provided by the full extent of the rhz ; habitability , where the rhz is truncated by the additional constraint of planetary orbital stability ( labelled as st and pt - type , respectively ) ; and cases of no habitability at all .
this classification scheme can be applied to both equal - mass and non - equal mass binary systems , as well as to systems with binaries in elliptical orbits , which will be the focus of the forthcoming paper ii of this series . as part of the current study a significant array of results
are given for a notable range of main - sequence stars , which are of both observational and theoretical interest .
a key aspect of the proposed method is the introduction of a combined algebraic formalism for the assessment of both s - type and p - type habitability ; in particular , mathematical criteria are presented allowing to determine for which systems s - type and p - type rhzs are realized . in this regard ,
a priori choices about the presence of s - type and p - type rhzs are neither necessary nor possible as the existence of s - type as well as p - type rhzs is proliferated through well - defined mathematical conditions pertaining to the underlying fourth - order algebraic equation .
the coefficients of the polynomial are given by the binary separation distance ( @xmath143 ) , the solar system - based parameter for the limit of habitability ( @xmath17 ) , and the modified values for the luminosities ( @xmath62 , @xmath63 ) of the stellar binary components , referred to as recast stellar luminosities . regarding the binary system habitable zone , we consider conservative , general and extended zones of habitability , noting that their inner and outer limits are informed by previous solar system investigations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in our segment of applications
, we examined the existence of habitable s - type orbits for selected examples .
we found that regarding the rhzs , owing to the typically relatively large separation of the stellar components , the effect of the stellar secondary on the extents of the rhzs is usually very minor .
the secondary s main influence on circumstellar habitability consists in imposing restrictions regarding planetary orbital stability implemented as an upper stability limit around each stellar component , which often truncates or nullifies s - type planetary habitability . in the framework of our study
, we specifically considered the radiative ehz , which is most outwardly extented ( i.e. , up to 2.4 au in case of the sun ) .
it was found that this kind of zone is most affected by the limitation of planetary orbital stability as it is located closest to the secondary stellar component .
furthermore , we also examined the existence of habitable p - type orbits . in this case , relatively complicated scenarios emerge .
in general , it was found that the best prospects for circumbinary habitability emerge for ( 1 ) systems with stellar components of relatively high luminosities ( no surprise here ! ) , ( 2 ) systems where the stellar luminosities are relatively similar ( for main - sequence stars , as implied by their stellar masses ) , and ( 3 ) systems of relatively small binary separations .
if conditions ( 2 ) or ( 3 ) are not met , it may occur that the outer rhl is located inside of the inner rhl , thus nullifying the rhz irrespectively of planetary orbital stability considerations . on the other hand ,
an unequal intrabinary distribution of masses entails a lower limit of planetary orbital stability ( i.e. , positioned closer to the binary system ) thus implying an enhanced opportunity for circumbinary habitability .
however , this aspect is of lesser significance for most systems compared to the restrictions for the rhzs due to the imbalance given by the stellar luminosities . various applications in this study concern stars of masses between 0.75 and 1.5 @xmath199 .
this approach is motivated to unequivocally demonstrate the effects of stellar binarity on the extent and structure of circumstellar habitability , which is most pronounced for massive , i.e. , highly luminous stars .
nonetheless , most stars in binaries are expected to be low - mass stars , i.e. , stars of spectral types k and m , owing to the skewness of the galactic initial mass function ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
for example , we compared pairs of systems given by ( 1.0 @xmath199 , 1.0 @xmath199 ) and ( 1.5 @xmath199 , 0.5 @xmath199 ) . obviously , the overall luminosity is by far greatest in the ( 1.5 @xmath199 , 0.5 @xmath199 ) system following the mass luminosity relationship ,
i.e. , @xmath220 ( e.g. , * ? ? ?
however , this system is found to be the highly unfavorable for the facilitation of circumbinary habitability .
particularly , it is found that the p - type ghz in the ( 1.0 @xmath199 , 1.0 @xmath199 ) system extends to 0.91 au , whereas it extends only to 0.65 au in the ( 1.5 @xmath199 , 0.5 @xmath199 ) system .
furthermore , smaller spatial extents are identified for p - type chzs , as this type of hz is in highest jeopardy owing to the relative small @xmath165 bracket compared to the @xmath216 bracket for ghzs ( see table 2 ) .
in fact , a considerable number of systems do not offer chzs at all , which again is a consequence of the radiative imbalance in those systems . also , the nullification of chzs in binary systems is most likely to occur in systems of relatively large separation distance .
in contrast , the best opportunities for facilitating circumbinary habitability is given in the context of ehzs , as expected .
future work will deal with a significant augmentation of our method to other systems , including systems with binary components in elliptical orbits ( see paper ii ) .
this will allow us to compare applications of our method , including results for individual systems , to other findings in the literature .
we also expect our method to be applicable to general binary systems with main - sequence stars as well as to systems containing evolved stars ; this latter effort is motivated by observational evidence and supporting theoretical efforts indicating that planets are also able to exist around stars that have left the main - sequence ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
particularly , it is highly desirable to augment our method to systems of higher order , as motivated by the steady progress in theory as well as ongoing and future observational discoveries of exosolar planetary systems .
this work has been supported in part by the seti institute .
the author acknowledges comments by b. quarles and z. e. musielak as well as assistance with computer graphics by s. sato , s. satyal , and m. sosebee .
the paper also benefited from detailed comments by an anonymous referee .
this study made use of the software applications fortran@xmath115 , mathematica@xmath115 , and matlab@xmath115 .
the author anticipates the development of a black box code , called binhab , to be hosted at the university of texas at arlington , which will allow the assessment of habitability in binary systems based on the developed method .
haghighipour , n. , dvorak , r. , & pilat - lohinger , e. 2010 , in planets in binary systems , ed .
n. haghighipour , astrophysics and space science library , vol .
366 ( new york : springer science + business media ) , p. 285
lccccc f0 & 7178 & 1.62 & 6.255 & 1.145 & 1.60 + f2 & 6909 & 1.48 & 4.481 & 1.113 & 1.52 + f5 & 6528 & 1.40 & 3.196 & 1.072 & 1.40 + f8 & 6160 & 1.20 & 1.862 & 1.037 & 1.19 + g0 & 5943 & 1.12 & 1.405 & 1.019 & 1.05 + g2 & 5811 & 1.08 & 1.194 & 1.009 & 0.99 + g5 & 5657 & 0.95 & 0.830 & 0.997 & 0.91 + g8 & 5486 & 0.91 & 0.673 & 0.985 & 0.84 + k0 & 5282 & 0.83 & 0.481 & 0.971 & 0.79 + k2 & 5055 & 0.75 & 0.330 & 0.957 & 0.74 + k5 & 4487 & 0.64 & 0.149 & 0.926 & 0.67 + k8 & 4006 & 0.53 & 0.066 & 0.905 & 0.58 + m0 & 3850 & 0.48 & 0.045 & 0.900 & 0.51 + lccc 1 & 0.84 & ghz / ehz & runaway greenhouse effect + 2 & 0.95 & chz & start of water loss + 3 & 1.00 & ... & earth - equivalent position + 4 & 1.37 & chz & first co@xmath3 condensation + 5 & 1.67 & ghz & maximum greenhouse effect , no clouds + 6 & 2.40 & ehz & maximum greenhouse effect , 100% clouds + lcccccc f0 & 1.98 & 2.25 & 2.33 & 2.91 & 3.66 & 5.49 + f2 & 1.70 & 1.93 & 2.00 & 2.54 & 3.18 & 4.72 + f5 & 1.46 & 1.65 & 1.72 & 2.24 & 2.78 & 4.08 + f8 & 1.13 & 1.28 & 1.34 & 1.78 & 2.19 & 3.19 + g0 & 0.99 & 1.12 & 1.17 & 1.58 & 1.94 & 2.80 + g2 & 0.91 & 1.03 & 1.09 & 1.48 & 1.81 & 2.61 + g5 & 0.77 & 0.87 & 0.91 & 1.25 & 1.53 & 2.19 + g8 & 0.69 & 0.78 & 0.83 & 1.15 & 1.39 & 1.99 + k0 & 0.59 & 0.67 & 0.71 & 0.99 & 1.20 & 1.71 + k2 & 0.49 & 0.55 & 0.59 & 0.84 & 1.01 & 1.43 + k5 & 0.34 & 0.38 & 0.40 & 0.59 & 0.71 & 0.99 + k8 & 0.22 & 0.25 & 0.27 & 0.41 & 0.49 & 0.67 + m0 & 0.19 & 0.21 & 0.23 & 0.35 & 0.41 & 0.56 + lccccccc f0 & 7178 & 5.545 & 5.625 & 5.464 & 4.509 & 4.801 & 5.223 + f2 & 6909 & 4.075 & 4.121 & 4.027 & 3.445 & 3.621 & 3.873 + f5 & 6528 & 3.005 & 3.027 & 2.981 & 2.681 & 2.770 & 2.897 + f8 & 6160 & 1.802 & 1.809 & 1.794 & 1.692 & 1.722 & 1.764 + g0 & 5943 & 1.382 & 1.384 & 1.379 & 1.337 & 1.349 & 1.366 + g2 & 5811 & 1.185 & 1.186 & 1.184 & 1.168 & 1.172 & 1.179 + g5 & 5657 & 0.832 & 0.832 & 0.832 & 0.837 & 0.836 & 0.834 + g8 & 5486 & 0.683 & 0.682 & 0.684 & 0.703 & 0.697 & 0.690 + k0 & 5282 & 0.494 & 0.493 & 0.496 & 0.523 & 0.515 & 0.505 + k2 & 5055 & 0.343 & 0.341 & 0.345 & 0.374 & 0.366 & 0.354 + k5 & 4487 & 0.150 & 0.149 & 0.151 & 0.177 & 0.170 & 0.161 + k8 & 4006 & 0.071 & 0.071 & 0.072 & 0.089 & 0.085 & 0.079 + m0 & 3850 & 0.050 & 0.049 & 0.051 & 0.064 & 0.060 & 0.055 + lccccc 1.50 & f2 v & 6842 & 4.233 & 3.830 & 1.96 + 1.25 & f7 v & 6256 & 2.173 & 2.078 & 1.44 + 1.00 & g2 v & 5833 & 1.228 & 1.216 & 1.10 + 0.75 & k2 v & 5100 & 0.357 & 0.372 & 0.61 + 0.50 & m0 v & 3857 & 0.045 & 0.050 & 0.22 + lcccc + 1 & 0.84 & 1.29 & 1.38 & 0.92 + 2 & 0.95 & 1.46 & 1.54 & 0.92 + 3 & 1.00 & 1.54 & 1.62 & 0.92 + 4 & 1.37 & 2.10 & 2.16 & 0.92 + 5 & 1.67 & 2.58 & 2.62 & 0.92 + 6 & 2.40 & 3.72 & 3.76 & 0.92 + + 1 & 0.84 & 1.19 & 1.52 & 0.79 + 2 & 0.95 & 1.36 & 1.69 & 0.79 + 3 & 1.00 & 1.45 & 1.79 & 0.79 + 4 & 1.37 & 1.96 & 2.28 & 0.79 + 5 & 1.67 & 2.43 & 2.76 & 0.79 + 6 & 2.40 & 3.58 & 3.91 & 0.79 + + 1 & 0.84 & 1.42 & 1.91 & 0.66 + 2 & 0.95 & 1.65 & 2.14 & 0.66 + 3 & 1.00 & 1.73 & 2.22 & 0.66 + 4 & 1.37 & 2.28 & 2.76 & 0.66 + 5 & 1.67 & 2.90 & 3.38 & 0.66 + 6 & 2.40 & 4.40 & 4.89 & 0.66 + lcccc + 1 & 0.84 & 1.21 & 1.54 & 1.83 + 2 & 0.95 & 1.40 & 1.69 & 1.83 + 3 & 1.00 & 1.48 & 1.76 & 1.83 + 4 & 1.37 & 2.06 & 2.28 & 1.83 + 5 & 1.67 & 2.54 & 2.72 & 1.83 + 6 & 2.40 & 3.70 & 3.83 & 1.83 + + 1 & 0.84 & 1.11 & 1.75 & 1.57 + 2 & 0.95 & 1.29 & 1.92 & 1.57 + 3 & 1.00 & 1.38 & 2.02 & 1.57 + 4 & 1.37 & 1.90 & 2.49 & 1.57 + 5 & 1.67 & 2.35 & 2.96 & 1.57 + 6 & 2.40 & 3.46 & 4.12 & 1.57 + + 1 & 0.84 & 1.21 & 2.16 & 1.32 + 2 & 0.95 & 1.43 & 2.38 & 1.32 + 3 & 1.00 & 1.51 & 2.47 & 1.32 + 4 & 1.37 & 2.05 & 3.00 & 1.32 + 5 & 1.67 & 2.66 & 3.62 & 1.32 + 6 & 2.40 & 4.17 & 5.13 & 1.32 + lcccc + 1 & 0.84 & 0.85 & 1.98 & 3.66 + 2 & 0.95 & 1.10 & 2.11 & 3.66 + 3 & 1.00 & 1.20 & 2.18 & 3.66 + 4 & 1.37 & 1.87 & 2.64 & 3.66 + 5 & 1.67 & 2.39 & 3.05 & 3.66 + 6 & 2.40 & 3.60 & 4.09 & 3.66 + + 1 & 0.84 & 0.70 & 2.23 & 3.15 + 2 & 0.95 & 0.95 & 2.40 & 3.15 + 3 & 1.00 & 1.07 & 2.50 & 3.15 + 4 & 1.37 & 1.69 & 2.95 & 3.15 + 5 & 1.67 & 2.18 & 3.42 & 3.15 + 6 & 2.40 & 3.32 & 4.55 & 3.15 + + 1 & 0.84 & 0.83 & 2.66 & 2.63 + 2 & 0.95 & 1.09 & 2.88 & 2.63 + 3 & 1.00 & 1.18 & 2.96 & 2.63 + 4 & 1.37 & 1.74 & 3.50 & 2.63 + 5 & 1.67 & 2.28 & 4.12 & 2.63 + 6 & 2.40 & 3.71 & 5.63 & 2.63 + lccccc + 1.5 & 2.47 & 1.90 & 1.50 & 0.87 & 0.65 + 1.25 & ... & 2.01 & 1.61 & 0.96 & 0.58 + 1.0 & ... & ... & 1.64 & 1.01 & 0.53 + 0.75 & ... & ... & ... & 1.00 & 0.47 + 0.5 & ... & ... & ... & ... & 0.43 + + 1.5 & 4.25 & 3.48 & 2.95 & 2.14 & 1.55 + 1.25 & ... & 3.26 & 2.74 & 1.93 & 1.27 + 1.0 & ... & ... & 2.57 & 1.77 & 1.10 + 0.75 & ... & ... & ... & 1.50 & 0.84 + 0.5 & ... & ... & ... & ... & 0.60 + + 1.5 & 7.38 & 7.15 & 5.96 & 4.43 & 3.29 + 1.25 & ... & 5.50 & 5.48 & 3.80 & 2.64 + 1.0 & ... & ... & 4.25 & 3.41 & 2.19 + 0.75 & ... & ... & ... & 2.40 & 1.55 + 0.5 & ... & ... & ... & ... & 0.92 + lcccccccccccccccccccc + 1.5 & 1.63 & 1.86 & 13.89 & 18.31 & 1.64 & 1.72 & 13.01 & 17.18 & 1.50 & ...
@xmath221 & 12.11 & 16.02 & 0.87 & ...
@xmath221 & 11.16 & 14.79 & 0.65 & ...
@xmath221 & 10.19 & 13.50 + 1.25 & ... & ... & ... & ... & 1.19 & 1.42 & 10.17 & 14.01 & 1.23 & 1.35 & 9.41 & 12.98 & 0.96 & ...
@xmath221 & 8.59 & 11.88 & 0.58 & ...
@xmath221 & 7.75 & 10.73 + 1.0 & ... & ... & ... & ... & ... & ... & ... & ... & 0.91 & 1.12 & 7.75 & 11.01 & 0.93 & 0.95 & 7.00 & 9.98 & 0.53 & ...
@xmath221 & 6.22 & 8.89 + 0.75 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & 0.50 & 0.65 & 4.26 & 6.37 & 0.47 & ...
@xmath221 & 3.70 & 5.57 + 0.5 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & 0.18 & 0.26 & 1.55 & 2.53 + + 1.5 & 1.43 & 2.32 & 12.22 & 22.85 & 1.44 & 2.14 & 11.44 & 21.44 & 1.53 & 2.06 & 10.65 & 19.99 & 1.70 & 1.87 & 9.82 & 18.45 & 1.55 & ...
@xmath221 & 8.96 & 16.85 + 1.25 & ... & ... & ... & ... & 1.05 & 1.75 & 8.97 & 17.27 & 1.08 & 1.65 & 8.30 & 15.99 & 1.16 & 1.45 & 7.58 & 14.64 & 1.27 & ...
@xmath221 & 6.83 & 13.22 + 1.0 & ... & ... & ... & ... & ... & ... & ... & ... & 0.80 & 1.36 & 6.85 & 13.46 & 0.82 & 1.16 & 6.18 & 12.19 & 0.94 & 1.01 & 5.50 & 10.86 + 0.75 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & 0.44 & 0.78 & 3.77 & 7.69 & 0.46 & 0.60 & 3.28 & 6.72 + 0.5 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & 0.16 & 0.30 & 1.38 & 3.00 + + 1.5 & 1.43 & 3.44 & 12.22 & 33.89 & 1.44 & 3.14 & 11.44 & 31.80 & 1.53 & 3.02 & 10.65 & 29.64 & 1.70 & 2.75 & 9.82 & 27.36 & 2.03 & 2.63 & 8.96 & 24.99 + 1.25 & ... & ... & ... & ... & 1.05 & 2.55 & 8.97 & 25.18 & 1.08 & 2.40 & 8.30 & 23.32 & 1.16 & 2.10 & 7.58 & 21.34 & 1.37 & 1.92 & 6.83 & 19.28 + 1.0 & ... & ... & ... & ... & ... & ... & ... & ... & 0.80 & 1.97 & 6.85 & 19.41 & 0.82 & 1.66 & 6.18 & 17.58 & 0.94 & 1.45 & 5.50 & 15.66 + 0.75 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & 0.44 & 1.10 & 3.77 & 10.90 & 0.46 & 0.85 & 3.28 & 9.51 + 0.5 & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & ... & 0.16 & 0.42 & 1.38 & 4.14 + | a comprehensive approach is provided to the study of both s - type and p - type habitability in stellar binary systems , which in principle can also be expanded to systems of higher order .
p - type orbits occur when the planet orbits both binary components , whereas in case of s - type orbits the planet orbits only one of the binary components with the second component considered a perturbator .
the selected approach encapsulates a variety of different aspects , which include : ( 1 ) the consideration of a joint constraint including orbital stability and a habitable region for a putative system planet through the stellar radiative energy fluxes ( radiative habitable zone " ; rhz ) needs to be met .
( 2 ) the treatment of conservative , general and extended zones of habitability for the various systems as defined for the solar system and beyond . ( 3 ) the providing of a combined formalism for the assessment of both s - type and p - type habitability ; in particular , mathematical criteria are presented for which kind of system s - type and p - type habitability is realized .
( 4 ) applications of the attained theoretical approach to standard ( theoretical ) main - sequence stars . in principle ,
five different cases of habitability are identified , which are : s - type and p - type habitability provided by the full extent of the rhzs ; habitability , where the rhzs are truncated by the additional constraint of planetary orbital stability ( referred to as st and pt - type , respectively ) ; and cases of no habitability at all .
regarding the treatment of planetary orbital stability , we utilize the formulae of holman & wiegert ( 1999 ) [ aj 117 , 621 ] as also used in previous studies .
in this work we focus on binary systems in circular orbits .
future applications will also consider binary systems in elliptical orbits and provide thorough comparisons to other methods and results given in the literature . |
the lie brackets of nilpotent lie algebras have some natural combinatorial structures which make this class of algebras very appealing ; both from theoretical and computational aspects .
the main purpose of this paper is to reconcile some of the theoretical nature of harmonic analysis on nilpotent lie groups with some of the computations involved in the process of constructing unitary irreducible representations .
it is worth noticing that pedersen has written several programs in reduce to compute polarizing subalgebras and canonical coordinates for all nilpotent lie algebras of dimensions less than seven .
in fact , michel duflo sent us copies of a companion manuscript to @xcite which contains outputs of his programs . since pedersen s programs were written a while ago , it is appropriate that we reintroduce some of his ideas through some more current technology .
let @xmath0 be a real nilpotent lie algebra .
we denote by @xmath1 the connected simply connected lie group corresponding to @xmath0 such that @xmath2 in this case , @xmath1 is called a nilpotent lie group .
the problem of classification of the irreducible unitary representations of @xmath1 is well understood @xcite .
let @xmath3 be a subalgebra of @xmath0 and let @xmath4 .
we say that @xmath5 is subordinated to @xmath6 if @xmath7 = 0.$ ] as a result , the formula @xmath8 defines a unitary character on @xmath9 .
the induced representation @xmath10 is an irreducible representation acting in the hilbert space @xmath11 if and only if @xmath12 is a * maximal isotropic subspace * for the skew - symmetric bilinear form @xmath13 defined by @xmath14 .$ ] such algebra is called a * polarizing subalgebra * @xcite for the linear functional @xmath6 and , it is well - known that the condition that @xmath12 is a maximal isotropic subalgebra is equivalent to 1 .
@xmath15 \subset\ker\left ( \ell\right ) $ ] 2 .
@xmath16 where @xmath17 the notation @xmath18 is the coadjoint action of @xmath1 on @xmath19 which is defined as follows:@xmath20 one of the complications of the representation theory of nilpotent lie groups is that the algebra @xmath12 is not generally unique , and there is no canonical way of constructing it .
however , there exist well - known procedures among which vergne s construction ( @xcite theorem @xmath21 is probably the most known .
the work presented in this paper is two - fold . in the second section of this paper ,
we revisit vergne s construction and we show that under specific conditions , the description of vergne is refinable . in the third section , we adapt our results to the class of finite - dimensional free nilpotent lie algebras of step - two .
we are then able to derive very simple descriptions of vergne polarizing subalgebras for all linear functionals in a dense open subset of @xmath22 at the end of the paper , we also provide a program written in mathematica to compute the polarizing subalgebra of any given nilpotent lie algebra .
let @xmath23 be a fixed strong malcev basis @xcite for the lie algebra @xmath0 passing through the center of @xmath24 let @xmath25 be a chain of ideals in @xmath0 such that @xmath26 given @xmath27 let @xmath28 where @xmath29 is the restriction of the linear functional @xmath6 to the vector space @xmath30 then , the polarizing subalgebra subordinated to the linear functional @xmath6 is described as follows @xmath31 where @xmath32 is the radical of the skew - symmetric bilinear form @xmath33 by definition @xmath34 a direct consequence of ( [ rad ] ) is that @xmath35 where @xmath36 is a skew - symmetric matrix of order @xmath37 given by @xmath38{cccc}0 & \ell_{j}\left ( \left [ z_{1},z_{2}\right ] \right ) & \cdots & \ell _ { j}\left ( \left [ z_{1},z_{j}\right ] \right ) \\ \ell_{j}\left ( \left [ z_{2},z_{1}\right ] \right ) & 0 & \cdots & \ell _ { j}\left ( \left [ z_{2},z_{j}\right ] \right ) \\ \vdots & \vdots & \ddots & \vdots\\ \ell_{j}\left ( \left [ z_{j-1},z_{1}\right ] \right ) & \ell_{j}\left ( \left [ z_{j-1},z_{2}\right ] \right ) & \cdots & \ell_{j}\left ( \left [ z_{j-1},z_{j}\right ] \right ) \\ \ell_{j}\left ( \left [ z_{j},z_{1}\right ] \right ) & \ell_{j}\left ( \left [ z_{j},z_{2}\right ] \right ) & \cdots & 0 \end{array } \right ] .\ ] ] it is worth mentioning that for lie algebras of arbitrarily large dimensions , it is a very difficult and tedious task to compute formula [ rad ] by hands .
therefore , there is a need to provide methods for the construction of vergne polarizing subalgebras that can be implemented in computer algebra systems and on other computational platforms . from a computational point of view ,
formula [ pol ] is slightly more appealing because of the use of linear algebra terms .
moreover , we should point out that the null - space of @xmath39 is obtained with respect to the fixed strong malcev basis @xmath40 for the lie algebra @xmath24 let us now provide an algorithm for the construction of a vergne polarizing subalgebra subordinated to an arbitrarily linear functional @xmath41 [ alg]let @xmath42 first , we set @xmath43{ccccc}0 & \ell\left [ z_{1},z_{2}\right ] & \cdots & \ell\left [ z_{1},z_{n-1}\right ] & \ell\left [ z_{1},z_{n}\right ] \\ -\ell\left [ z_{1},z_{2}\right ] & 0 & \cdots & \ell\left [ z_{2},z_{n-1}\right ] & \ell\left [ z_{2},z_{n}\right ] \\ \vdots & \vdots & \ddots & \vdots & \vdots\\ -\ell\left [ z_{1},z_{n-1}\right ] & -\ell\left [ z_{2},z_{n-1}\right ] & \cdots & 0 & \ell\left [ z_{n-1},z_{n}\right ] \\ -\ell\left [ z_{1},z_{n}\right ] & -\ell\left [ z_{2},z_{n}\right ] & \cdots & -\ell\left [ z_{n-1},z_{n}\right ] & 0 \end{array } \right ] \label{matrixm}\ ] ] such that @xmath44 is a singular skew - symmetric matrix of order @xmath45 next , we define a submatrix of @xmath46 of order @xmath37 as follows : @xmath47{ccccc}0 & \ell\left [ z_{1},z_{2}\right ] & \cdots & \ell\left [ z_{1},z_{j-1}\right ] & \ell\left [ z_{1},z_{j}\right ] \\ -\ell\left [ z_{1},z_{2}\right ] & 0 & \cdots & \ell\left [ z_{2},z_{j-1}\right ] & \ell\left [ z_{2},z_{j}\right ] \\ \vdots & \vdots & \ddots & \vdots & \vdots\\ -\ell\left [ z_{1},z_{j-1}\right ] & -\ell\left [ z_{2},z_{j-1}\right ] & \cdots & 0 & \ell\left [ z_{j-1},z_{j}\right ] \\ -\ell\left [ z_{1},z_{j}\right ] & -\ell\left [ z_{2},z_{j}\right ] & \cdots & -\ell\left [ z_{j-1},z_{j}\right ] & 0 \end{array } \right ] .\ ] ] finally @xmath48 although , it is fairly easy to implement algorithm [ alg ] , it is not generally the most efficient way to compute the algebra @xmath49 in the remainder of this section , we will provide a refined version of algorithm [ alg ] .
let @xmath50 be the central ideal for the lie algebra @xmath24 if @xmath51 then @xmath52 this lemma follows from the fact that if we assume that @xmath53 then the matrix @xmath54 is simply a zero matrix of order @xmath37 which we consider as a linear operator acting on the vector space @xmath55 thus its null - space is equal to the whole vector space @xmath30 now , let us assume that @xmath56 then the matrix @xmath36 is equal to@xmath57{ccccccc}0 & \cdots & 0 & 0 & 0 & \cdots & 0\\ \vdots & & \vdots & \vdots & \vdots & & \vdots\\ 0 & \cdots & 0 & 0 & 0 & \cdots & 0\\ 0 & \cdots & 0 & 0 & \ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2}\right ] & \cdots & \ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{j}\right ] \\ 0 & \cdots & 0 & -\ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2}\right ] & 0 & & \ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2},z_{j}\right ] \\ \vdots & & \vdots & \vdots & \vdots & \ddots & \vdots\\ 0 & \cdots & 0 & -\ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{j}\right ] & -\ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2},z_{j}\right ] & \cdots & 0 \end{array } \right ] .\label{me}\ ] ] we will consider the following submatrix of @xmath36 which we denote by @xmath58{cccc}0 & \ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2}\right ] & \cdots & \ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{j}\right ] \\ -\ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2}\right ] & 0 & & \ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2},z_{j}\right ] \\ \vdots & \vdots & \ddots & \vdots\\ -\ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 1},z_{j}\right ] & -\ell_{j}\left [ z_{\dim\mathfrak{z}\left ( \mathfrak{g}\right ) + 2},z_{j}\right ] & \cdots & 0 \end{array } \right]\ ] ] where @xmath59@xmath38{cc}0 & 0\\ 0 & m_{0}\left ( \ell_{j}\right ) \end{array } \right ] \label{block}\ ] ] and @xmath60 we observe that if @xmath61 then @xmath62 is a submatrix of @xmath63 it is fairly easy to check that the above implies that if @xmath64 then @xmath65 now , let @xmath66 clearly , it is possible for the set @xmath67 to be empty .
however , in the case where @xmath67 is nonempty , it makes sense to attempt to refine formula [ pol ] . if @xmath68 then @xmath69 if @xmath68 then @xmath70 is a skew - symmetric matrix of even full - rank .
thus , its null - space is the trivial vector space .
therefore , @xmath71 appealing to the above lemma , and to formula [ pol ] , the following is immediate .
[ t1]let @xmath6 be a linear functional in @xmath22 a vergne polarizing subalgebra subordinated to the linear functional @xmath6 is @xmath72
in this section , we will adapt the algorithms provided in the previous section to a class of nilpotent lie algebras known as free nilpotent lie algebras of step - two . applying
the refined algorithms described in theorem [ t1 ] , we are able to provide very simple descriptions of vergne polarizing algebras subordinated to a family of linear functionals in a zariski open ( dense ) subset of the linear dual of the corresponding lie algebra .
let @xmath0 be the free nilpotent lie algebra of step - two on @xmath73 generators @xmath74 . if @xmath75 is the generating set for @xmath0 then @xmath76 such that @xmath77 the non - trivial lie brackets of this lie algebra are described as follows .
@xmath78 = z_{ij\text { } } \text{for } \left ( 1\leq i\leq m\text { and } i < j\leq m\right ) .\ ] ] it is then easy to see that @xmath79 in these settings , we define recursively the matrix @xmath58{ccccc}0 & \ell_{j}\left [ z_{1},z_{2}\right ] & \cdots & \ell_{j}\left [ z_{1},z_{j-1}\right ] & \ell_{j}\left [ z_{1},z_{j}\right ] \\ \ell_{j}\left [ z_{2},z_{1}\right ] & 0 & \cdots & \ell_{j}\left [ z_{2},z_{j-1}\right ] & \ell_{j}\left [ z_{2},z_{j}\right ] \\ \vdots & \vdots & \ddots & \vdots & \vdots\\ \ell_{j}\left [ z_{j-1},z_{1}\right ] & \ell_{j}\left [ z_{j-1},z_{2}\right ] & \cdots & 0 & \ell_{j}\left [ z_{j-1},z_{j}\right ] \\ \ell_{j}\left [ z_{j},z_{1}\right ] & \ell_{j}\left [ z_{j},z_{2}\right ] & \cdots & \ell_{j}\left [ z_{j},z_{j-1}\right ] & 0 \end{array } \right]\ ] ] such that @xmath58{cc}m_{0}\left ( \ell_{j-1}\right ) & v\left ( \ell_{j}\right ) \\ w\left ( \ell_{j}\right ) & 0 \end{array } \right]\ ] ] where @xmath80{c}\ell_{j}\left [ z_{1},z_{j}\right ] \\ \ell_{j}\left [ z_{2},z_{j}\right ]
\ell_{j}\left [ z_{j-1},z_{j}\right ] \end{array } \right ] \text { and } w\left ( \ell_{j}\right ) = \left [ \begin{array } [ c]{cccc}\ell_{j}\left [ z_{j},z_{1}\right ] & \ell_{j}\left [ z_{j},z_{2}\right ] & \cdots & \ell_{j}\left [ z_{j},z_{j-1}\right ] \end{array } \right ] .\ ] ] from now on , we will assume that @xmath4 and @xmath81 let @xmath82 be a natural number smaller than the dimension of the lie algebra @xmath83 furthermore , for each @xmath82 we define the embedding map @xmath84 such that @xmath85 is a map which sends the column vector @xmath86^{t}$ ] to the vector @xmath87 and @xmath86^{t}$ ] is the transpose of @xmath86.$ ] [ l1]if @xmath88 is odd then the null - space of the matrix @xmath89 is equal to @xmath90{cccc}0 & \ell_{j}\left [ z_{1},z_{2}\right ] & \cdots & \ell_{j}\left [ z_{1},z_{j-1}\right ] \\
\ell_{j}\left [ z_{2},z_{1}\right ] & 0 & \cdots & \ell_{j}\left [ z_{2},z_{j-1}\right ] \\
\vdots & \vdots & \ddots & \vdots\\ \ell_{j}\left [ z_{j-1},z_{1}\right ] & \ell_{j}\left [ z_{j-1},z_{2}\right ] & \cdots & 0 \end{array}\right ] ^{-1}\left [ \begin{array } [ c]{c}\ell_{j}\left [ z_{1},z_{j}\right ] \\ \ell_{j}\left [ z_{2},z_{j}\right ] \\
\vdots\\ \ell_{j}\left [ z_{j-1},z_{j}\right ] \end{array } \right ] \right ) \right ) .\ ] ] if @xmath37 is even then @xmath91 for any linear functional @xmath6 satisfying ( [ zariski ] ) ) , we will show that the set @xmath92 only contains even indices .
first of all , let us suppose that @xmath37 is odd .
then @xmath93 is a skew - symmetric matrix of even * full*-rank ( see ( [ zariski ] ) ) .
so @xmath57{cc}m_{0}\left ( \ell_{j-1}\right ) ^{-1 } & 0\\ 0 & 0 \end{array } \right ] \left [ \begin{array } [ c]{cc}m_{0}\left ( \ell_{j-1}\right ) & v\left ( \ell_{j}\right ) \\ w\left ( \ell_{j}\right ) & 0 \end{array } \right ] = \left [ \begin{array } [ c]{cc}i & m_{0}\left ( \ell_{j-1}\right ) ^{-1}v\left ( \ell_{j}\right ) \\ 0 & 0 \end{array } \right]\ ] ] and@xmath57{cc}i & m_{0}\left ( \ell_{j-1}\right ) ^{-1}v\left ( \ell_{j}\right ) \\ 0 & 0 \end{array } \right]\ ] ] is just the row - reduced form of the matrix @xmath57{cc}m_{0}\left ( \ell_{j-1}\right ) & v\left ( \ell_{j}\right ) \\
w\left ( \ell_{j}\right ) & 0 \end{array } \right ] .\ ] ] therefore , @xmath94 and the above is precisely equal to @xmath90{cccc}0 & \ell_{j}\left [ z_{1},z_{2}\right ] & \cdots & \ell_{j}\left [ z_{1},z_{j-1}\right ] \\
\ell_{j}\left [ z_{2},z_{1}\right ] & 0 & \cdots & \ell_{j}\left [ z_{2},z_{j-1}\right ] \\ \vdots & \vdots & \ddots & \vdots\\ \ell_{j}\left [ z_{j-1},z_{1}\right ] & \ell_{j}\left [ z_{j-1},z_{2}\right ] & \cdots & 0 \end{array}\right ] ^{-1}\left [ \begin{array } [ c]{c}\ell_{j}\left [ z_{1},z_{j}\right ] \\ \ell_{j}\left [ z_{2},z_{j}\right ] \\
\vdots\\ \ell_{j}\left [ z_{j-1},z_{j}\right ] \end{array } \right ] \right ) \right ) .\ ] ] secondly , by assumption , if @xmath37 is even then @xmath95 is a skew - symmetric matrix of even full - rank .
therefore the null - space of this matrix is trivial .
if @xmath73 is odd ( @xmath96 for some natural number @xmath97 ) then@xmath98 if @xmath73 is even ( @xmath99 for some natural number @xmath97 ) then @xmath100 the above proposition is a direct application of lemma [ l1 ] and theorem [ t1 ] .
let @xmath0 be a real nilpotent lie algebra of dimension @xmath101 with a fixed strong malcev basis : @xmath102 we will present a program written in mathematica which can be used to compute the vergne polarizing subalgebra with respect to a linear functional @xmath103 .
the only argument for this program is the matrix @xmath104 the output of this program is a spanning set for a polarizing subalgebra subordinated to @xmath105 here are some actual mathematica outputs
we thank michel duflo for bringing to our attention that niels pedersen has already written programs in reduce to compute polarizing subalgebras for all nilpotent lie algebras of dimensions less than seven , and we also thank him for supplying reference @xcite .
g. folland , a course in abstract harmonic analysis , studies in advanced mathematics .
crc press , boca raton , fl , @xmath110 .
n. pedersen , geometric quantization and the universal enveloping algebra of a nilpotent lie group .
( 1989 ) , no .
2 , 511563.1088 - 6850 | according to the orbit method , the construction of a unitary irreducible representation of a nilpotent lie group requires a precise computation of some polarizing subalgebra subordinated to a linear functional in the linear dual of the corresponding lie algebra .
this important step is generally challenging from a computational viewpoint . in this paper
, we provide an algorithmic approach to the construction of the well - known vergne polarizing subalgebras @xcite .
the algorithms presented in this paper are specifically designed so that they can be implemented in computer algebra systems .
we also show there are instances where vergne s construction could be refined for the sake of efficiency .
finally , we adapt our refined procedure to free nilpotent finite - dimensional lie algebras of step - two to obtain simple and precise descriptions of vergne polarizing algebras corresponding to all linear functionals in a dense open subset of the linear dual of the corresponding lie algebra .
also , a program written for mathematica is presented at the end of the paper . |
atlas is a general purpose detector in construction at the @xmath0 collider @xmath1 at cern .
it is designed to exploit the full discovery potential as given by center of mass energy of @xmath2 tev and the highest luminosity of more than @xmath3 .
the optimization of the detector is driven by the requirements to detect new particles and physics processes as e.g. the higgs boson in the decay @xmath4 , heavy @xmath5and @xmath6like particles or supersymmetric particles . in the atlas detector the hadronic liquid argon ( lar ) end - cap calorimeter ( hec ) @xcite covers the pseudorapidity range @xmath7 . to fully exploit the physics potential of atlas , an energy resolution for jets of typically @xmath8
is required for the hec .
the desired linearity of the energy response measured has to stay within @xmath9 @xcite .
the total energy containment up to the highest energies as well as an acceptable low background in the muon chambers require a calorimeter thickness of at least @xmath10 interaction lengths ( @xmath11 ) including the electromagnetic end - cap calorimeter ( emec ) in front of the hec .
fast electronics is needed in order to keep the pile - up low .
liquid argon technology was chosen as the active medium for its robustness against the high radiation levels present in this forward region . in the following the status of the construction of the hec
is discussed and the most recent results from the @xmath12 beam test measurements are presented .
the lar hec @xcite is a sampling calorimeter with flat copper absorber plates .
it shares the two end - cap cryostats together with the emec and forward calorimeter ( fcal ) .
the hec is structured in two wheels , the front hec@xmath13 and the rear hec@xmath14 wheel , placed in the cryostat behind the emec wheel .
each wheel has an outer diameter of about @xmath15 .
the length of hec@xmath13 ( hec@xmath14 ) is @xmath16 ( @xmath17 ) .
the thickness of the copper absorber plates is @xmath18 for hec@xmath13 and @xmath19 for hec@xmath14 , with the first plate being half of this normal thickness in either case .
each wheel is made out of @xmath20 identical modules .
the weight of hec@xmath13 is @xmath21 and that of hec@xmath14 is @xmath22 . in total @xmath23 gaps for hec@xmath13 and @xmath24 gaps for hec@xmath14
are instrumented with a read - out structure .
longitudinally they are read out as segments of @xmath25 and @xmath24 gaps for hec@xmath13 and @xmath25 and @xmath25 gaps for hec@xmath14 .
the read - out structure is based on the principle of an electrostatic transformer ( est ) @xcite .
the size of the liquid argon gaps between the electrode structure is @xmath26 .
the transverse granularity , driven by the aim to reconstruct the decay @xmath27 at high @xmath28 , is @xmath29 for the region @xmath30 and @xmath31 beyond @xmath32 .
the hec employs the concept of `` active pads '' : the signals from individual pads are fed into separate preamplifiers ( based on highly integrated gaas electronics ) and summed actively . the electronics boards with these ic s are positioned directly at the module periphery , the chips are operated in the liquid argon .
thus the input capacitances are minimized and a short signal rise time ensured .
the use of cryogenic gaas preamplifiers and summing amplifiers provides the optimum signal - to - noise ratio for the hec .
the `` active pad '' electronics has been employed in more than @xmath33 cold tests with and without beam .
it has proven to be very reliable .
the construction of the hec is almost finished : all the modules are assembled and have passed the cold tests , all four wheels are assembled and one full end - cap with hec@xmath13 and hec@xmath14 wheels is fully integrated into the end - cap cryostat together with the emec and the fcal .
the level of high voltage ( hv ) problems after integration is @xmath34 , not causing any acceptance losses , but rather a few specific corrections for some channels . only one out of @xmath35 read - out channels is not operational , corresponding to a failure rate of @xmath36 .
the @xmath12 beam test has been carried out in the h6 beam - line at the cern sps providing hadrons , electrons or muons in the energy range @xmath37 .
the load in the liquid argon cryostat consisted of one emec module ( @xmath38 of the full emec wheel ) , three hec@xmath13 modules ( @xmath39 of the full hec@xmath13 wheel ) and two hec@xmath14 modules .
constrained by the cryostat dimension the depth of the hec@xmath14 modules were half of the atlas modules .
the impact angle of beam particles was @xmath40 with respect to the front plane , yielding a non - pointing geometry of the setup in @xmath41 ( vertically ) unlike the atlas situation . in front of the first emec layer a presampler end - qcap module
was placed inside the cryostat .
it allowed studies of preshower corrections with optional additional material in front of the cryostat .
finally , scintillators for triggering and timing and @xmath42 mwpcs with horizontal and vertical wire planes for beam position reconstruction were present further upstream in the beam line .
in the analysis of the beam test data , the signal reconstruction follows closely the methods used in previous stand - alone beam tests of the emec @xcite and the hec @xcite .
the output signals of the cold hec summing amplifiers and the raw signals from the emec were carried to the front - end boards ( feb ) outside the cryostat . here
the amplification of the emec signals and signal shaping of all signals was performed .
the signals were sampled at a rate of @xmath43 and stored in a switched capacity array of the feb . for each event @xmath44 ( @xmath24 )
samples per emec ( hec ) channel were recorded together with the mwpc response , trigger information and the tdc measured delay between the trigger and the @xmath43 sampling clock .
the raw adc samples were processed with an optimal filtering ( of ) method @xcite using @xmath45 event samples . for the hec
the detailed knowledge of each component in the electronics chain and the form of the input calibration pulse was used to determine the response function , which was then used to predict the shape for the physics signals .
the resulting predicted physics shapes together with the autocorrelation matrices from noise runs was used for the computation of the of weights . unlike the final situation in atlas with its fixed delay between trigger and sampling clock
the beam test trigger was asynchronous with respect to the sampling clock .
the of weights were therefore calculated in steps of @xmath46 in order to fill the @xmath47 trigger window and parametrized by a @xmath48 order polynomial .
the accuracy for the amplitude reconstruction following this method achieved the level of @xmath49 . the required conversion factor from adc counts to na
was obtained by a proper study of calibration pulse hight .
the total signal deposited in the calorimeter was estimated on event by event basis using a cell - based two - dimensional topological clustering algorithm .
the electronic noise @xmath50 ( in na ) for each cell was obtained by studying either muon data or time samples located outside the physics pulse time region .
each cluster consisted of at least one cell with a signal - to - noise ratio above 4 ( @xmath51 ) .
a threshold on the absolute value of the signal - to - noise ratio , @xmath52 , was applied to all other cells .
they were included in the cluster if they shared at least one edge with a cluster member cell satisfying @xmath53 .
the symmetric cuts on the cell and neighbor level avoided biases due to electronics noise .
two super - clusters for the emec and the hec were defined by summing all cluster signals in the emec and the hec respectively . for the hec the signals in the @xmath54 layer were multiplied by @xmath14 in order to account for the @xmath55 smaller sampling fraction .
from monte carlo simulations of the beam test configuration the leakage outside the emec was found to be very small for electrons and has been neglected . therefore the ratio of the known beam energy ( @xmath56 ) and the sum of all signals in the emec in na defined the electromagnetic scale factor , @xmath57 , where the error is here statistical only . the variation with energy tests the linearity in the energy range considered and was found to be better than @xmath58 .
the energy resolution for electrons has been studied using the super - cluster mentioned above . in data
the energy resolution was found to be @xmath59 after noise subtraction .
the noise @xmath60 varies with energy due to the non - fixed cluster size .
geant3 @xcite and geant4 @xcite ) based monte carlo simulations are in good agreement with the data and yield for geant3 an energy resolution @xmath61 while geant4 yield a @xmath62 .
further tuning is required for atlas to improve the fair agreement with the data .
the electromagnetic scale for the hec was taken from the previous stand - alone beam test @xcite , @xmath63 , taking the modified electronics into account . a good agreement of the total visible energy in the emec and hec for pions with monte carlo simulations based either on geant3 or the quark - string - gluon - plasma ( qgsp ) model of geant4
is observed , while the geant4 low - and - high - energy - pion - parameterization ( lhep ) model deviate largely from data .
the non - compensating nature of the two calorimeters makes weighting of hadronic energy deposits necessary . a cell based weighting method which was successfully used in previous experiments
@xcite needs a detailed simulation on the cell level , which is not yet available for atlas .
therefore a more coarse weighting scheme on the super - cluster level has been applied . with the leakage outside the detector volume as predicted by the monte carlo and the known beam energy 6 weights ( 3 for the emec and 3 for the hec )
have been fitted from the two super - cluster energies and their energy density , leading to the weighted energies @xmath64+c_3\right)$ ] .
is slightly worse than expected from monte carlo .
the geant3 simulation predicts a @xmath66 and a geant4 lhep and qgsp simulations predict an energy resolution sampling term of @xmath67 and @xmath68 , respectively . in general the geant4 models
seem closer to the data , but neither qgsp and lhep give an optimal description . the different energy dependence of the geant4 predictions is also reflected in non - vanishing constant terms : @xmath69 for lhep and @xmath70 for qgsp . the ratio of the combined weighted energy of emec and hec over the combined electromagnetic energy yield the effective @xmath71-ratio for the end - cap calorimeters ranging from @xmath72 at @xmath73 to @xmath74 at @xmath75 for pions .
[ fig : eoverpi ] shows this ratio for the data and the different mc models .
the energy dependence is in all cases rather similar .
however the mc predictions are here substantially below the data , with geant3 being especially low .
9 the atlas collaboration , atlas liquid argon calorimeter technical design report , cern - lhcc-96 - 041 , atlas tdr 2,15 december 1996 .
the atlas collaboration , atlas calorimeter performance design report , cern - lhcc-96 - 040 , atlas tdr-1 , 13 january 1997 .
j. colas , m. pripstein , w.a .
wenzel , nucl .
inst . and meth . a 294 ( 1990 ) 583 .
b. aubert et al .
( atlas electromagnetic liquid argon calorimeter collab . ) , nucl . inst . and meth . a 500 ( 2003 ) 178 b. dowler et al .
( atlas liquid argon hadronic end - cap calorimeter collab . ) , nucl .
inst . and meth . a 482 ( 2002 ) 94
w. e. cleland and e. g. stern , nucl .
inst . and meth . a 338 ( 1994 ) 467 .
r. brun et al . , geant3 , cern - dd / ee/84 - 1 ( 1986 ) a. kiryuning and d. salihagi , monte carlo for the hec prototype : software and example of analysis , atlas hec note-063 ( 1998 ) .
s. agostinelli et al . , geant4 - a simulation toolkit , nucl .
a 506 ( 2003 ) 250 .
d. dowler et al .
( atlas liquid argon hec collab . ) , nucl .
inst . and meth . a 482 ( 2002 ) 94
. i. abt et al .
( h1 collab . ) , nucl . inst . and meth . a 386 ( 1997 ) 348 .
b. andrieu et al .
( h1 calorimeter collab . ) , nucl .
inst . and meth . a 336 ( 1993 ) 499 . | atlas has chosen for its hadronic end - cap calorimeter ( hec ) the copper - liquid argon sampling technique with flat plate geometry and gaas pre - amplifiers in the argon .
the contruction of the calorimeter is now approaching completion .
results of production quality checks are reported and their anticipated impact on calorimeter performance discussed .
selected results , such as linearity , electron and pion energy resolution , uniformity of energy response , obtained in beam tests both of the hadronic end - cap calorimeter by itself , and in the atlas configuration where the hec is in combination with the electromagnetic end - cap calorimeter ( emec ) are described . |
quasars whose optical emission is hidden by dust in the optical and gas in the soft x - ray have always been hard to find . however
, a large type-2 ( hidden ) quasar population has long been predicted , both from the statistics of seyfert-1 and 2 galaxies , and from those of radio galaxies and radio - loud quasars . a large population of hidden agn is also predicted from models of the x - ray background ( e.g. comastri et al .
1995 ; worsley et al .
2005 ) , though most models predict these agn to be of seyfert rather than quasar luminosity .
some success in finding dust obscured quasars has been obtained using selection based on the two micron all sky survey ( 2mass ) to find lightly dust - reddened ( rest frame extinctions @xmath3 ) quasars , based either on 2mass colors alone ( cutri et al .
2001 ) , or on 2mass combined with a radio detection and very red optical - ir colors ( glikman et al .
only recently , however , have large samples of type-2 quasars , with rest - frame @xmath4 been found .
zakamsaka et al .
( 2003 ) used the sloan digital sky survey ( sdss ) to find objects with the high - ionization narrow - line spectra characteristic of type-2 agn . combining _ iso _ and 2mass data allowed haas et al .
( 2004a ) and leipski et al .
( 2005 ) to find significant numbers of obscured agn , including several type-2 quasars .
_ spitzer _ data has been used to find highly - obscured agn , both through mid - infrared color selection ( lacy et al .
2004 [ hereafter paper i ] , 2005a ; stern et al . 2005 ; alonso - herrero et al . 2005
; polletta et al . 2006 ) and joint radio - infrared selection ( martnez - sansigre et al .
2005 , 2006 ; weedman et al . 2006 ; donley et al .
type-1 quasars selected through joint spitzer and optical colors and/or morphologies have been targeted by several groups ( brown et al .
2005 ; hatziminaoglou et al . 2005 ; siana et al .
also , the mid - infrared properties of x - ray selected agn have been discussed by treister et al .
( 2005 ) and barmby et al .
. mid - infrared selection of agn is effective because dust heated by the agn forms a `` calorimeter '' , whose luminosity is not as strongly affected by obscuration as that in most other wavelength bands ( except the radio ) ( e.g. spinoglio et al .
1995 ; meisenheimer et al .
2001 ; ogle et al .
richards et al .
( 2006 ) show that the bolometric correction in the mid - infrared varies little for different subsamples of type-1 quasars ( e.g. optically or infrared luminous , optically - red or optically blue ) , or luminosity , and that about 15% of the bolometric luminosity of a quasar is emitted at mid - infrared wavelengths .
integration over the richards et al .
( 2006 ) mean quasar sed as a whole suggests that about 40% of the emission from a typical quasar is reprocessed by dust in the infrared .
this means that mid - infrared selection of agn is probably a good proxy for selection on the basis of bolometric luminosity , at least for those objects which are not so highly obscured that their mid - infrared luminosity is significantly supressed .
much of the dust emission is thought to arise in the `` obscuring torus '' required by unified schemes ( e.g. antonucci 1993 ) , though it is becoming increasingly clear that host dust in the narrow line region of agn also contributes significantly ( e.g. mason et al .
agn are an important global contributor to the mid - infrared luminosity density .
brand et al . ( 2006 ) show that the luminosity density of mid - ir emission in samples of mips 24@xmath0 m sources selected at mjy levels is dominated by agn , only at submjy levels do star forming galaxies begin to dominate . in this paper
, we focus on optical spectroscopic follow - up of agn candidates selected according to the mid - infrared color criteria presented in paper i. by selecting relatively bright objects from wide - area _
surveys we are able to find many obscured agn with mid - infrared luminosities typical of quasars at moderate redshifts .
these objects provide a good complement to the mostly lower - luminosity or higher redshift agn found in surveys such as the great observatories origins deep survey ( goods ) and the chandra deep fields .
preliminary results from the spectroscopic follow - up of the sample presented in paper 1 were given in lacy et al .
( 2005a ) . in this paper
, we describe the selection of samples of mid - infrared selected agn candidates flux - limited at 24@xmath0 m from both the extragalactic first look survey ( xfls ) and the _ spitzer _ wide - area infrared extragalactic survey ( swire ) xmm - lss field ( lonsdale et al .
2005 ) , and their spectrosopic follow - up observations .
we classify the optical spectra according to the type of agn , and use them to place a lower bound on the fraction of obscured quasars and luminous seyfert galaxies .
we also investigate the x - ray fluxes of those objects which fall on archival observations made with the x - ray multiple mirror telescope ( _ xmm - newton _ ) . a more detailed analysis of the spectral energy distributions ( seds ) and x - ray properties of the agn will be carried out in a later paper ( petric et al .
2006 ) .
we assume a cosmology with @xmath5 , @xmath6 and @xmath7 .
we selected samples of agn by mid - infrared color from the xfls and swire xmm fields .
the basis of our color selection technique is described in sajina , lacy & scott ( 2005 ) and paper i , though the samples in this paper are flux - limited at 24@xmath0 m rather than at 8@xmath0 m as was the case for the sample of paper 1 . to obtain the xfls sample , we began by matching the xfls 24@xmath0 m catalog of fadda et al .
( 2006 ) to the infrared array camera ( irac ) 4-band catalog of lacy et al .
all mips 24@xmath0 m detections with detections in all four irac bands were then plotted on the color - color diagram of figure 1 . for the swire xmm sample
, we used the bandmerged swire irac / mips 24@xmath0 m catalog available from the _ spitzer _ popular products website .
the color cut used to select agn candidates was @xmath8 and @xmath9 and @xmath10 . the @xmath11 cut was moved by @xmath12 compared to paper i. the exact position of this cut is not critical , and it was felt that shifting this would help remove non - agn contaminants , particularly as the final 5.8@xmath0 m fluxes in the xfls catalogue were increased by @xmath13% in the final version of the xfls catalog relative to the catalog used in paper i. we discuss the exact placement of the color cuts with respect to the completeness of the agn selection in section 5.3 .
candidates were selected down to limiting 24@xmath0 m fluxes of 4.4mjy in the xfls and 6.6mjy in the swire xmm field .
the 24@xmath0 m flux limited sample should be an improvement on 8@xmath0 m selection by making us more sensitive to highly - reddened objects , and high redshift objects , where the strong @xmath14-correction on the mid - infrared dust emission means that they drop out of flux - limited samples selected at shorter wavelengths .
the disadvantage of this selection is that we will be missing objects at @xmath15 which have strong silicate absorption in the 24@xmath0 m band , however , as our median redshift is @xmath16 only a few objects will be missed due to this effect .
the xfls sample is complete for objects falling in the irac `` agn wedge '' color selection to a flux limit of 4.6mjy at @xmath17 m , and the swire sample to a flux limit of 6.6mjy ( note that all mips sources down to these flux limits have 4-band irac detections ) .
approximately 20% of mips 24@xmath0 m sources at these flux levels fall into our agn - dominated part of the color - color plot .
a few stars and nearby galaxies have been removed from the samples .
most stars were only present in the initial selection due to photometric errors related to saturation in the shorter irac bands , though one star ( sstxfls j172432.8 + 592646 , see paper i ) has the correct colors to fall in the sample , presumably due to a dust disk .
very low redshift ( @xmath18 ) galaxies are also possible contaminants due to the presence of 6.2@xmath0 m pah emission in the 5.8@xmath0 m band of irac ( sajina , lacy & scott 2005 ) .
the interacting galaxy arp 54 appeared in the selection of the swire agn sample , but has been removed for this reason .
alonso - herrero , a. et al .
2006 , apj , 640 , 167 antonucci , r. 1993 , ara&a , 31 , 473 baldwin , j.a . phillips , m.m . & terlevich , r. 1981 , pasp , 93 , 5 barmby , p. et al .
2006 , apj in press ( astro - ph/0512618 ) becker , r.h .
2001 , apjs , 135 , 227 brand , k. et al .
2006 , apj , in press ( astro - ph/0602198 ) brown , m.j .
et al . 2006 ,
apj , 638 , 88 caccianiga , a. , et al .
2004 , , 416 , 901 chiappetti , l. et al .
2005 , a&a , 439 , 413 comastri , a. , setti , g. , zamorani , g. & hasinger , g. 1995 , a&a , 296 , 1 comastri , a. et al . 2002 ,
apj , 571 , 771 condon , j.j . , cotton , w.d .
, yin , q.f . , shupe , d.l .
, storrie - lombardi , l.j . ,
helou , g. , soifer , b.t . & werner , m.w .
2003 , aj , 125 , 2411 g.p .
2003 , apj , 584 , l57 cutri , r.m . ,
nelson , b.o . ,
kirkpatrick , j.d .
, huchra , j.p . & smith , p.s .
2001 , in asp conf .
232 , the new era of wide field astronomy , ed .
r. clowes , a. adamson , & g. bromage ( san francisco : asp ) , 78 donley , j.l . ,
rieke , g.h . , rigby j.r . &
perez - gonzales , p.g .
2005 , apj , 634 , 169 eckart , m.e .
2006 , apjs , 165 , 19 fadda , d. , jannuzi , b. , ford , a. & storrie - lombardi , l.j .
2004 , aj , 128 , 1 fadda , d. et al .
2006 , aj , 131 , 2859 glikman , e. , gregg , m.d . ,
lacy , m. , helfand , d.j .
, becker , r.h . & white , r.l .
2004 , apj , 607 , 60 grimes , j.p .
kriss , g.a . & espey , b.r . 1999 ,
apj , 526 , 130 haas , m. et al .
2004a , a&a , 219 , l49 haas , m. et al . 2004b , a&a , 424 , 531 hatziminaoglou , e. 2005 , aj , 129 , 1198 kauffmann , g. et al .
2003 , mnras , 346 , 1055 kewley , l.j . , groves , b. , kauffmann , g. & heckman , t. 2006 , mnras , submitted ( astro - ph/0605681 ) lonsdale , c.j .
2003 , pasp , 115 , 897 lacy , m. et al .
2004 , apjs , 154 , 166 lacy , m. et al . 2005a , mmsai , 76 , 154 lacy , m. et al .
2005b , apjs,161 .
41 leipski , c. et al .
2005 , a&a , 440 , l5 lawrence , a. 1991 , mnras , 252 , 586 martnez - sansigre , a. et al .
2005 , nat , 436 , 666 martnez - sansigre , a. et al .
2006 , mnras , 370 , 1479 mason , r.e .
2006 , apj , 640 , 612 meisenheimer , k. , haas , m. , mller , s.a.h . ,
chini , r. , klaas , u. & lemke , d. 2001 , a&a , 372 , 719 michel , l. , herent , o. motch , c. , pye , j. & watson .
2004 , in , astronomical data analysis software and systems xiii , oschenbein , m. , allen , m. & egret , d. eds , asp conf .
314 , p. 570
moran , e.c . ,
filippenko , a.v .
& chornock , r. 2002 , apj , 579 , 71 petric , a. et al .
2006 , in preparation norman , c. 2002 , apj , 571 , 218 ogle , p. , whysong , d. & antonucci , r. 2006 , apj , submitted ( astro - ph/0601485 ) papovich , c. et al .
2006 , aj , 132 , 231 pei , y.c .
1992 , apj , 395 , 130 polletta , m. et al .
2006 , apj , 642 , 673 rayner , j.t .
2003 , pasp , 115 , 362 richards , g.t .
2006 , apjs , submitted ( astro - ph/0601558 ) rigby , j.r . ,
rieke , g.h . ,
donley , j.l .
, alonso - herrero , a. & prez - gonzlez , p.g .
2006 , apj , 645 , 115 siana , b. et al .
2006 , to appear in `` infrared diagnostics of galaxy evolution '' , ed .
r. chary ( astro - ph/0604373 ) sajina , a. , lacy , m. & scott , d. 2005 , apj , 621 , 256 sharp , r. g. , sabbey , c. n. , vivas , a. k. , oemler , a. , mcmahon , r. g. , hodgkin , s. t. , & coppi , p. s. 2002 , , 337 , 1153 & yang , t. 2003 , apj , 596 , l23 soltan , a. 1982 , mnras , 200 , 115 spinoglio , l. , malkan , m.a .
, rush , b. , carrasco , l. & recillas - cruz , e. 1995 , apj , 453 , 616 treister , e. et al .
2004 , apj , 616 , 123 trump , j.r .
2006 , apjs , in press ( astro - ph/0606016 ) van ojik , r. , rttgering , h.j.a . , miley , g.k . ,
bremer , m.n .
, maccetto , f. & chambers , k.c . 1994 , a&a , 289 , 54 watson , m.g . , et al .
2004 , aas / high energy astrophysics division , 8.3701 weedman , d. le floch , e. , higdon , s.j.u . ,
higdon , j.l . & houck , j.r .
2006 , apj , 638 , 613 worsley , m.a .
2005 , mnras , 357 , 1281 yu , q. & tremaine , s. 2002 , mnras , 335 , 965 yuan , f. & narayan , r. 2004 , apj , 612 , 724 zakamska , n.l .
2003 , aj , 126 , 2125 | we present optical spectroscopy of a sample of 77 luminous agn and quasars selected on the basis of their mid - infrared colors .
our objects are selected from the _ spitzer _ extragalactic first look survey and swire xmm - lss fields , with a typical 24@xmath0 m flux density of 5mjy .
the median redshift is 0.6 , with a range of @xmath1 .
only 33% ( 25/77 ) of these objects are normal type-1 quasars , with no obscuration .
44% ( 34/77 ) are type-2 objects , with high - ionization , narrow emission lines , 14% ( 11/77 ) are dust - reddened type-1 quasars , showing broad lines but a dust - reddened or unusually weak quasar continuum .
9% ( 7/77 ) show no sign of an agn in the optical spectrum , having either starburst spectra or spectra which could be of either starburst or liner types .
these latter objects are analogous to the x - ray detected population of agn with weak or non - existent optical agn emission ( the `` xbongs '' ) .
21 of our objects from the swire field fall within moderately - deep xmm exposures .
all the unobscured quasars , and about half the obscured quasars are detected in these exposures .
this sample , when taken together with other samples of _ spitzer _ selected agn and quasars , and results from x - ray studies , confirms that obscured agn dominate the agn and quasar number counts of all rapidly - accreting supermassive black hole systems , at least for @xmath2 .
this implies a high radiative efficiency for the black hole accretion process . |
the main goal of this paper is the numerical research of the effect of the soliton type disturbance of the temperature in the point of the contact of the free boundaries .
this effect we consider in the case of the phase field model for the stefan - gibbs - thomson problem .
the difference between this problem and the classical stefan problem is that the surface tension is taken into account in the stefan - gibbs - thomson problem . at the beginning of the paper
we briefly give the main analytic results about the confluence of the free boundaries and then we illustrate these results by computer simulation . the effect of the soliton type ( negative ) disturbance of the temperature @xmath0 of the free boundaries in the point of the contact is shown in fig .
[ f0 ] .
we present the results of the computer simulation for the phase field system in the spherically symmetric case .
namely , we consider the domain @xmath1 $ ] , where @xmath2 $ ] is the spherical layer in the spherical coordinates .
we assume that the domain @xmath3 is divided into the three layers @xmath4 and @xmath5 as follows : @xmath6 , \quad \omega^{-}(t)=[r_{1}(t),r_{2}(t ) ] , \quad \omega^{+}_{2}(t)=[r_{2}(t ) , r_{2}],\ ] ] where @xmath7 , @xmath8 is the free boundaries between phases `` @xmath9 '' and `` @xmath10 '' .
we assume that the phase `` @xmath9 '' occupies the layer @xmath4 , and the phase `` @xmath10 '' occupies the layer @xmath5 ( see fig . [ f1 ] ) . ) .
a ) coordinate and time dependence of the temperature .
b ) soliton type disturbance of the temperature in the neighborhood of the point of the contact of the free boundaries .
c ) initial data of the temperature and the evolution of the free boundaries @xmath11 , @xmath12 in prime . ] in this case the phase field system have the form @xcite @xmath13 where @xmath14 , @xmath15 , \quad t\in[0,t_{1}].\ ] ] the function @xmath16 has the meaning of the temperature , and the function @xmath17 ( so - called the order function ) determines the phase state of the medium .
namely , the value @xmath18 corresponds to the phase `` @xmath10 '' in the layer @xmath5 , and the value @xmath19 corresponds to the the phase `` @xmath9 '' in the layers @xmath20 .
we denote @xmath21 the phase field system ( [ l:1 ] ) we can write in the form @xmath22 passing to the limit as @xmath23 in ( [ phheat ] ) , ( [ phac ] ) we obtain the stefan - gibbs - thomson problem for the each free boundary ( see @xcite ) @xmath24,\quad r\neq \hat{r}_{i}(t),\quad i=1,2,\label{sh:1}\\ & \left.\frac{\overline{\sigma}}{r}\right|_{r=\hat{r}_{i}(t)}=(-1)^{i+1}\left ( \hat{r}_{i}'(t)+\frac{2}{\hat{r}_{i}(t)}\right),\label{sgt:1}\\ & \left.\left[\frac{\partial\overline{\sigma}}{\partial r}\right ] \right|_{r=\hat{r}_{i}(t)}=(-1)^{i+1}2\hat{r}_{i}(t)\hat{r}'_{i}(t).\label{sst:1}\end{aligned}\ ] ] this passage to the limit is possible , for example , in the case where the corresponding limit problems have classical solutions . in this case , the weak limits as @xmath23 of solutions ( [ phheat ] ) , ( [ phac ] ) give these solutions @xcite .
the existence of the classical solution @xmath25 of problem ( [ sh:1])([sst:1 ] ) is discussed in @xcite . by @xmath26
we denote the instant of time of the confluence of the free boundaries , and @xmath27 is the sphere of the contact of the free boundaries .
the smooth approximations ( i.e. the approximate solution of the system ( [ phheat ] ) , ( [ phac ] ) with misalignment that is small in the weak sense as @xmath23 , see @xcite ) of the stefan - gibbs - thomson problem ( of the phase field system ) is constructed in @xcite in the one dimensional case and this smooth approximations exists in the time included the instant of the contact @xmath28 . moreover , this smooth approximations is constructed on the assumption of the existence of the classical solution of the limit problem as @xmath29 , where @xmath30 is any number .
this approximations admits the passage to the limit as @xmath23 .
it is possible to show @xcite that in the common sense the asymptotic solution of the system ( [ phheat ] ) , ( [ phac ] ) has the form @xmath31.\notag\end{aligned}\ ] ] as @xmath29 , @xmath32 . here @xmath33 as @xmath34 , @xmath35 as @xmath36 , @xmath37 , @xmath8 is the smooth functions , @xmath38 , @xmath39 , and @xmath40;\,\mathbb{s}(\mathbb{r}^{2}_{z}))$ ] . by @xmath41
we denote the schwartz space of smooth rapidly decreases functions .
if the initial data for ( [ phheat ] ) , ( [ phac ] ) has the form ( [ as:1 ] ) , ( [ as:2 ] ) at @xmath42 , then , for @xmath29 , we have the estimate @xmath43 where @xmath44 is a solution of system ( [ phheat]),([phac ] ) ( see @xcite and references in ) . here
$ ] , and the constant @xmath46 is independent of @xmath47 .
the main obstacle to the construction of approximations of solution in the case of confluence of free boundaries is the fact that , instead of an ordinary differential equation whose solution is the function @xmath48 @xcite , in the case of confluence of free boundaries , we must deal with a partial differential equation for which the explicit form of the exact solution is unknown . in papers @xcite using the the weak asymptotic method the solution of the phase field system
was constructed that describes the confluence of the free boundaries .
namely , the ansatz of the order function has the form .
@xmath49\notag.\end{aligned}\ ] ] the temperature is sought in the form @xmath50 @xmath51)$ ] , @xmath52 for @xmath53 $ ] . here
@xmath54 is the model of the temperature , i.e. it is the function of the simplest structure which describes the behaviors of the temperature qualitatively correct , and @xmath55 is an unknown smooth function .
namely , @xmath56 here @xmath57 @xmath58 so , from the given above formulas we see that the model @xmath54 is linear in @xmath59 in the layers @xmath60 , @xmath8 and parabolic in @xmath59 in the layer @xmath5 . in paper
@xcite the formulas are obtained those determine the functions contained in ansatzes ( [ au:1 ] ) , ( [ new23 ] ) in the one dimensional case .
the analysis of these formulas shows that the contained in the phase field system temperature @xmath61 has the soliton type disturbance in the neighborhood of the point of the contact .
this disturbance is localized in the space coordinates and in time .
the `` width '' of this localization is proportional to @xmath47 . in @xcite
the limit as @xmath23 of the amplitude of the disturbance is derived . in the one dimensional case the value of this amplitude
is given by formula @xmath62\right|_{x
= x^*,\ t = t^ * } = -\lim_{t\to t^{*}-0}\lim_{\varepsilon\to 0}\frac{v_{1}+v_{2}}{2},\ ] ] where @xmath63 and @xmath64 are the velocities of the merging ( one - dimensional ) free boundaries .
the analytic treatment based on the weak asymptotics method shows that such effect is general .
for example , the exactly same formula is correct in the case of merging globe layers @xcite those is shown in fig.[f1 ] .
@xmath62\right|_{r = r^*,\ t = t^ * } = -\lim_{t\to t^{*}-0}\lim_{\varepsilon\to 0}\frac{v_{1n}+v_{2n}}{2},\ ] ] @xmath65 and @xmath66 are the normal velocities of the merging three - dimensional free boundaries ( correspondingly , @xmath67 and @xmath68 ) . if the free boundaries are asymmetrical , then their principle curvatures have the like signs at the point of the contact . in this case
the amplitude of the jump of the temperature is determined by formula @xmath62\right|_{(x , y , z)=(x^*,y^*,z^*),\ t = t^ * } = -\lim_{t\to t^{*}-0}\lim_{\varepsilon\to0}\left(\frac{|v_{1n}|+|v_{2n}|}{2}-| \mathcal{k}_1- \mathcal{k}_2|\right)\ ] ] in the instant of the confluence and at the point of the contact .
here @xmath69 , @xmath70 is the point of the contact of the free boundaries , and @xmath71 , @xmath72 are the principle curvatures of the free boundaries at the point of the contact .
let us consider another situation .
we assume that the solution of our problem is symmetric about @xmath73-axe ( see fig.[f2 ] ) .
by @xmath74 and @xmath75 we denote the points in which the free boundaries cross the @xmath73-axe and these points is situated on the shorts distance one from another .
we assume that @xmath76 is the point of the contact of the free boundaries .
the principle difference between this situation and the situation in fig.[f1 ] is that in the instant of the contact @xmath77 the principle curvatures of the free boundaries have the unlike signs at the point of the contact @xmath78 . in this case
the amplitude of the jump of the temperature is determined by formula @xmath62\right|_{(x , y , z)=(0,y^*,0),\ t = t^ * } = -\lim_{t\to t^{*}-0}\lim_{\varepsilon\to 0 } \left(\frac{|v_{1n}|+|v_{2n}|}{2}+|\mathcal{k}_1|+|\mathcal{k}_2|\right).\ ] ] the mentioned soliton type disturbance is observed in the numerical experiments , see fig.[f : tminjump ] , [ f : tt ] .
we note that beside the our own papers we know only the single paper of a.m. meirmanov and b.a .
zaltsman @xcite . in this paper
the problem of the confluence of the free boundaries in the case of the hele - shaw problem is derived and this problem is the special case of the stefan - gibbs thomson problem .
the regularized ( not limit ) problem is considered in @xcite and in this case the effect of the soliton type disturbance is not shown .
the analog of this fact is the following problem .
let us consider the heat equation in a rectilinear segment and with not zero ( @xmath79 ) dirichlet s initial condition .
clearly , the solution of this problem is zero .
however , the numerical solution is not identical zero for the different scheme with a node in the point @xmath80 .
the choice of the different scheme for system ( [ phheat ] ) , ( [ phac ] ) is based on some ideas that are sufficiently general for solving nonlinear equations .
namely , we calculate the heat equation ( [ phheat ] ) at the next time step and , consequently split system ( [ phheat ] ) , ( [ phac ] ) .
we associate eq .
( [ phac ] ) with implicit different scheme @xmath81 + \varkappa \sum_{i=1}^{n}\frac{\sigma^{k}_{i}}{r_{i}}.\notag\end{aligned}\ ] ] here @xmath82 is the mesh order function at the @xmath83-th time step , @xmath84 is the mesh `` temperature '' at the @xmath83-th time step . the first term ( the term in the square brackets ) in the right hand part of eq .
( [ meshu ] ) is obtained as following .
we denote @xmath85 we linearize the mesh function @xmath86 as following @xmath87 system ( [ meshu ] ) is completed with initial and boundary conditions in the points @xmath88 , @xmath89 .
it is clear that eqs .
( [ meshu ] ) is the three - point equations relatively to @xmath90 and these equations are solved by the sweep method . to calculate the mesh `` temperature '' @xmath91 we use the standard different scheme for the heat equation ( [ phheat ] ) .
namely , @xmath92 equations ( [ mesht ] ) is also solved by the sweep method .
we use the main segment @xmath93 $ ] ( i.e. , @xmath94 , @xmath95 ) for the numerical simulation of the process of the confluence . for the different values of @xmath47 : 1 .
@xmath96 , 2 .
@xmath97 , 3 .
@xmath98 , 4 .
@xmath99 , 5 .
@xmath100.,width=264 ] the initial data we choose as provided by the structure of the weak asymptotic solution ( [ as:2 ] ) .
we use the function @xmath101 as the initial data for the order function @xmath102 . here
@xmath103 , @xmath104 determine the initial position of the free boundaries ( see fig.[f : u0],[f : t0 ] ) .
the initial data @xmath105 for the `` temperature '' is taken in the form in fig.[f : t0 ] as provided by model ( [ modt:1 ] ) .
namely , the function @xmath105 is parabolic in the domain with phase `` @xmath10 '' ( between the free boundaries ) , and the function @xmath105 is linear in the domain with phase `` @xmath9 '' . at the same time , the dependence between the initial positions and the initial velocities of the free boundaries and the values of @xmath105 is determined by formula ( [ sgt:1])in the boundary points @xmath106 , @xmath107 . .
the initial positions of the free boundaries are @xmath103 , @xmath104,width=264 ] it is clear that the considered initial data differ from the exact solution of the problem .
nevertheless it is known that the solutions of problem ( [ phheat ] ) , ( [ phac ] ) converge to solutions of the limit stefan - gibbs - thomson problem as @xmath23 and on the sufficiently general assumptions .
more other , the solution of the cauchi problem come to self - similar regime by the widely known behaviours of the semi - linear parabolic equations ( see @xcite ) . beside it , the formulas for the asymptotic solution are the deformations of the formulas for the semi - similar solution .
let us consider the results of the numerical simulation of the process of the confluence of the free boundaries .
some graphics given below illustrate the common behaviors of the solution of system ( [ phheat ] ) , ( [ phac ] ) and confirm the certainty of the numerical results .
we simulate for the different ( decreasing ) values of the parameter @xmath47 , @xmath96 , @xmath97 , @xmath98 , 4 .
@xmath99 , @xmath100 .
the mesh is equal @xmath108 and the time step is equal @xmath109 . in fig.[f :
tjump ] , [ f : ujump ] are shown the profiles of the temperature @xmath0 and the order function @xmath102 in the neighborhood of the point of the contact of the free boundaries ( for fixed @xmath110 ) for different values of the parameter @xmath47 . from the graphics in fig.[f : tjump ] we see that the `` width '' of the neighborhood of the instant of time of the jump of the temperature decrease ( it is proportional to @xmath47 ) owing to decreasing @xmath47 . at the same time
, the jump occurs at the the different instant of time for the different values of the parameter @xmath47 .
namely , in dependence of the decreasing of @xmath47 the sequence of the instant of time @xmath111 ( in which the jump becomes minimum ) increases , see also table [ t : tmin ] . beside it , from table [ t : tmin ] we see that the intervals grow short between the neighboring instants @xmath111 as the parameter @xmath47 decreases .
the cause of this fact is that the width of the transition zone of the order function @xmath102 is different for different values of @xmath47 , see formula ( [ u0 ] ) .
namely , the width of transition zone is proportional @xmath47 ( as larger the value of @xmath47 , as wider the transition zone ) , see fig .
[ f : u0 ] and @xcite . as a result of
this fact the contact of the transition zones ( and , consequently the beginning of the confluence of the free boundaries ) occurs previously for the simulation process with larger @xmath47 , see fig .
[ f : ur ] . clearly , the width of the confluence is also proportional to @xmath47 respect to @xmath59 . from fig .
[ f : ur ] we see that for the fixed instant of time @xmath112 the free boundaries corresponding to the graphic 1 ( @xmath96 ) complete the confluence , but at the same time the free boundaries corresponding another graphics ( for smaller @xmath47 ) are on the sufficiently large distance .
.coordinate and time dependence of the jump of the temperature @xmath0 . @xmath120 and @xmath111 are values of the coordinate @xmath59 and time @xmath121 at which the jump becomes minimum . [ cols="^,^,^,^",options="header " , ] in column @xmath122 $ ] the data for the jump are given obtained by formula ( [ sigma ] ) , and in column @xmath123 $ ] the data for the jump are given calculated by numerical simulation . from table [ t : tjump ] we see that the analytical data converge to the numerical data as @xmath47 decreases .
v.g.danilov , g.a.omelyanov & e.v.radkevich _ asymptotic solution of a phase field system and the modified stefan problem .
_ 1995 , differentialnie uravneniya 31(3 ) , 483491 ( in russian).(english translation in differential equations 31(3 ) , 1993 . | in this paper we numerically research the solutions of the phase field system for the spherically symmetric stefan - gibbs - thomson problem in the case of interaction of the free boundaries .
we analyze the effect of the soliton type disturbance of the temperature in the point of the contact of the free boundaries . |
tight scaling relations between the central bh mass and various properties of their host galaxies in the local universe ( velocity dispersion of the bulge component , @xmath9 , stellar masses , m@xmath10 , luminosity ) , have been discovered after the launch of the hubble space telescope ( hst ) nearly 20 years ago .
these correlations @xcite have revolutionized the way we conceive the physical link between galaxy and agn evolution , suggesting a possible physical coupling between bhs and host galaxies in their evolution . the local scaling relations , although important in establishing the role of smbhs in galaxy evolution , have been unable to uniquely determine the physical nature of the smbh - galaxy coupling .
a large number of theoretical models ( some including feedback ) have been proposed which can reproduce them reasonably well @xcite , but make different predictions for their redshift evolution .
constraining the high - z m@xmath7-m@xmath8 plane is thus of fundamental importance in discriminating between different models but it is unfortunately observationally complex .
first of all , we have to rely on less - accurate , single - epoch virial mass estimators based on broad agn emission lines .
moreover , in deriving the host galaxy properties , we are limited by the presence of the bright nucleus . at high redshift
, most studies of the m@xmath7-m@xmath10 relationship have focused on optically blue type1 broad line agn ( blagn ) ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) for which the uv broad lines ( e.g. mgii and civ , observed in the optical band at high - z ) are detected and virial formulas can be used to derive bh masses from the agn luminosity and the width of the broad lines ( see e.g. , * ? ? ?
* ; * ? ? ?
* for recent reviews ) . the emerging picture ( see * ?
* for a comprehensive picture ) points towards a mild or absent evolution of the average black - hole - to - host - galaxy - mass ratio up to z@xmath111 @xcite and a positive one for z@xmath121 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in particular , putting together all available observations of agn at [email protected] - 7.1 that have reliable estimates of m@xmath7 , of the stellar masses of the host galaxies , and of bulge - to - disk ratios ( for low - z objects only ) , @xcite finds that the m@xmath7/m@xmath13 evolves as ( 1+z)@xmath14 , with @xmath15=0.7 - 2 .
the above results , have been interpreted either as evidence that black holes form faster , at earlier epochs , than their host galaxies , or as due to an evolution of the intrinsic scatter of the m@xmath7-m@xmath10 relation ( or a combination of both effects ) . however , the sign of evolution seems to weaken at z@xmath16 when considering the total galaxy mass for galaxies that are not bulge - dominated @xcite . according to theoretical models ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the positive evolution found for unobscured bright agn at high redshift , can be interpreted as the fact that galaxy mergers , the driving mechanism for feeding bhs in bright qsos , are more common at higher redshift , and are thus able to increase the bh accretion rate more rapidly than the star - formation rate at such redshifts . on the contrary , at lower redshift ( z@xmath171 ) , for moderate - luminous , seyfert - like objects with lower bh masses living in spiral - galaxy , secular galaxy evolution , occurring mostly in galaxy disks , would increase the stellar masses , with black hole accretion proceeding at a lower rate . a radical alternative for explaining the m@xmath7-m@xmath18 relation emerges from the statistics of mass averaging in galaxy mergers .
this idea , first proposed by @xcite , has been elaborated using more realistic semi - analytic models of galaxy formation properly embedded in @xmath19cdm merger trees by @xcite . according to these simulations , the creation of the scaling relations
can be fully explained by the hierarchical assembly of bh and stellar mass through galaxy merging , from an initially uncorrelated distribution of bh and stellar masses in the early universe .
in contrast to the observational results mentioned above , @xcite found that for agn hosted in submm selected galaxies ( smgs ) at [email protected] , the growth of smbhs actually lags that of the host stellar mass .
the apparently contradictory behavior can be interpreted as a selection bias possibly affecting in the opposite way these studies .
in fact , studying luminous agn we are biased towards selecting the most massive smbhs ( see * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ) , while a sample of smgs will be biased towards massive stellar hosts ( see * ? ? ?
* ) since the galaxies that are undergoing the most intense star formation at z@xmath112 also appear to be massive galaxies at this epoch ( e.g. * ? ? ?
in addition to possible bias in the sample selection , quasars with different properties may represent different phases in black hole and galaxy evolution .
although smgs are very interesting high-@xmath21 ( @xmath22 ; e.g. , * ? ?
* ) objects with large star - formation activities ( up to few 1000 m@xmath23yr@xmath24 ; @xcite ) , they represent the most extreme class of sources of the entire star - forming galaxy population .
new important constraints in understanding the physical nature of the agn - galaxy coupling require the extension of the analysis of the bh - galaxy relation to the population of x - ray obscured , dust reddened qsos .
current agn - galaxy co - evolution models predict an early dust - enshrouded phase associated with rapid bh growth triggered by galaxy mergers @xcite .
tidal interactions trigger both violent star formation and funneling of large amount of gas into the nuclear region which feed and obscure the accreting smbh ( e.g. , * ? ? ?
* ; * ? ? ?
during this phase , we expect the central source to be buried by the surrounding material and the agn to likely appear as an optically type2 agn , x - ray obscured and red ( e.g. * ? ? ? * ; * ? ? ? * ) .
so far , studies on this population are restricted to very few peculiar objects and the statistic is therefore not enough to draw a global picture .
the observation of red qsos is in fact more challenging compared to the blue ones .
measurements of bh masses for red qsos at z@xmath121 can not be derived from optical spectroscopy : the uv broad emission lines , used to estimate the bh masses through virial estimators , are not detected due to dust extinction . for these sources
, we can detect the agn optical broad lines e.g. , h@xmath5 , redshifted in the near - ir , since dust extinction is reduced with respect to the rest - frame uv lines ( see * ? ? ?
until recently , there have been only a handful of published works on the evolution of the scaling relations between bhs and host masses for red qsos : the above mentioned work of @xcite based on a sample of 4 qsos detected in smgs , the work from @xcite based on three hard x - ray selected , obscured , red qsos with broad h@xmath5 emission line and the most recent work from @xcite based on a 2mass+first selected sample of 13 red qsos at 0.4@xmath17z@xmath171 .
@xcite found that the m@xmath7/m@xmath10 ratio at z=1 - 2 is in agreement with the local bh - galaxy relation , and lower than the average m@xmath7/m@xmath10 ratio observed in blue qsos in the same redshift range .
however , this result is still tentative since it is based on only three agn . on the contrary
, @xcite found that most of the sources in their sample lie below the local relation and that objects below the local relation are the one accreting at high eddington ratio ( @xmath250.3 ) .
here we present new ir spectra obtained with ` sinfoni ` and ` xshooter ` at vlt of 9 x - ray obscured , red qsos for which a broad h@xmath5 component has been observed in the ir spectra .
we complement these data with additional sources available in literature reaching a final total sample of 21 x - ray obscured , intermediate type @xcite , red qsos with broad h@xmath5 emission observed in the nir .
this represents the most extensive study so far for this class of targets .
the aim of the paper is to study the m@xmath7-m@xmath10 scaling relation and its evolution for this class of sources by comparing them with the results obtained for optically blue qsos .
the paper is organized as follows : in section [ sec : sample ] we will introduce our sample , starting from the new observations and proceeding with the samples taken from the literature and included in our analysis . in section [ sec : bh ] and [ sec : masses ] we derive virial bh masses and host galaxy stellar masses .
star formation rates and eddington ratios are presented in section [ sec : sfr_edd ] .
results on the m@xmath7-m@xmath10 plane are shown in section [ sec : scaling ] followed by the comparison with the semi - analytical model prediction in section [ sec : model ] .
finally , discussion and conclusions are presented in section [ sec : summary ] .
unless otherwise stated , uncertainties are quoted at the 68% ( 1@xmath9 ) confidence level .
this work is based on new near - ir observations ( program 88.b-0316(a ) with ` sinfoni ` and 090.a-0830(a ) with ` xshooter ` at the eso vlt ) of hard x - ray detected , obscured ( @xmath26 ) agn with red ( @xmath27 ) colors which show a broad h@xmath5 line , integrated with data collected from literature .
the final sample consists of 21 objects , 9 of them are new targets while the remaining 12 are data taken from the literature as listed below . the sample is summarized in table [ tab : sample ] .
our observations were performed with ` sinfoni ` and ` x - shooter ` at vlt . `
: ` we obtained near - ir spectra using ` sinfoni ` at vlt of eleven hard x - ray selected obscured ( @xmath28 ) red ( @xmath29 ) qsos at @xmath30 , selected from the _ chandra deep field south _
( cdfs ) ( 1ms exposure , * ? ? ? * ) .
the sources were already spectroscopically classified as type2 agn in the optical band and had unambiguous spectroscopic redshifts ( goods - s , * ? ? ?
as our sample consisted of sources at different redshifts , observations were obtained in the @xmath31 , @xmath32 and @xmath33 bands to detect the h@xmath5 line . the resolutions in the three bands are respectively @xmath34 , @xmath35 and @xmath36 . the requested on - source integration time was 2.3 hours .
however , of the 11 sources targeted , only 7 have been observed completely while 2 of them were observed for only 1.5 hours and the last 2 for 45 minutes .
the typical seeing conditions during the observations were 0.5@xmath37@xmath37 - 1.1@xmath37@xmath37 .
data have been reduced by using the eso - sinfoni pipeline .
this pipeline performs the background subtraction , flat - fielding , the spectral calibration of each individual slice , and then reconstructs the cube . out of these targets we detected a broad h@xmath5 component for 4 of them ( three of which
are observed for the whole requested time while cdfs633 only for 45 minutes ) .
here we focus on these sources for which the h@xmath5 broad component allowed us to infer their bh masses .
if we restrict the sample to z@xmath121 and @xmath38}>1\times10^{-15 } erg\ , cm^{-2}s^{-1}$ ] we have seven objects , 4 of which with broad h@xmath39 detected .
at the time of the observations , one source had a wrong redshift estimate and thus the h@xmath5 line could not be detected while , for another one , the source was not at the center of the observed field .
therefore , we can conclude that out of 5 hard x - ray selected with @xmath38}>1\times10^{-15 } erg\ , cm^{-2}s^{-1}$ ] , obscured ( @xmath28 ) red ( @xmath29 ) qsos at @xmath40 , we detect a broad h@xmath39 line in 4/5 of the sample . ` : ` nir spectra have been obtained using ` xshooter ` at vlt @xcite of 10 x - ray obscured and luminous agn at 1.25@[email protected] selected from the cosmos field ( i.e. xmm - cosmos sample , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the parent sample consisted of @xmath41 `` obscured qsos '' candidates , selected on the basis of their red optical colors ( @xmath42 ) and high x - ray to optical flux ratio and/or bright ir emission compared to the optical one ( see * ? ? ?
* sec . 7 for details ) . among them , the observed sample consisted of the 6 k - brightest sources in the @xmath43 redshift range with only photometric redshift available , plus 4 spectroscopically confirmed objects satisfying the same redshift and magnitude selection ( 3 narrow - line objects and 1 broad line agn ) .
compared to the cdfs sources , these sources are brighter thus extending the range of luminosities covered by our sample to higher values . the exposure times ranged from 1hr to 2hrs with typical seeing conditions of 0.5@xmath37@xmath37 - 1.0@xmath37@xmath37 .
spectra have been taken with a resolution of r=5100 ( 0.9 slit ) .
the reduction has been done using reflex 2.4 . a detailed description of the full sample , data reduction and analysis can be found in brusa et al .
( 2014 ) . here
we focus on 5 sources at @xmath44 erg / s and @xmath45 for which a broad component has been detected and hence bh masses could be computed .
one of these sources is cosmos2028 , whose vlt / sinfoni spectra have been already published by ( * ? ? ?
* ulasj1002 + 0137 ) .
given the lower resolution of their spectra ( r=1500 ) , @xcite could not deblend the narrow and broad h@xmath5 components . the measured fwhm and
consequently the bh mass is thus lower than the one obtained in this work . if we consider the z@xmath121 and @xmath38}>1\times10^{-14 } erg\ , cm^{-2}s^{-1}$ ]
sources , we find that 5/6 of the observed sources have h@xmath5 broad detected . from these two observational campaigns , we thus find that , selecting hard x - ray obscured agn with red colors , we are able to build a highly complete sample of objects ( @xmath4680% ) with broad h@xmath5 emission lines , above a certain flux .
overall we have obtained a final sample of 9 sources with broad h@xmath5 component .
we fitted the line profile as explained in sec .
[ sec : fwhm ] and derived the full width at half maximum of the line ( fwhm ) .
stellar masses have been computed with a 2-component sed fitting procedure as explained in sec .
[ sec : masses ] . in order to increase the statistic of the study we complemented the new data with a collection of high - z sources taken from literature with similar properties .
following is the list of included samples : + * _ 3 qsos from @xcite _ + we include in our sample three x - ray obscured , red ( @xmath47 ) qsos from @xcite .
ir spectra of hard x - ray sources detected in the hellas2xmm @xcite and elais - s1 survey @xcite have been taken with ` sinfoni ` and ` isaac ` , and a broad h@xmath5 line has been detected in three cases .
this sample is the most similar to our new observations . for these objects ,
we re - analyzed the spectra after improving the reduction , sky subtraction and extraction .
6 qsos hosted in smgs from @xcite _ + we include in our work a sample of 6 smgs that host x - ray identified obscured agn with broad h@xmath5 line from @xcite .
four of these sources ( the one with available stellar masses measured from @xcite ) were already reported in @xcite .
here we recomputed the stellar masses for all of the 6 sources using the sed fitting method as explained in sec .
[ sec : masses ] after collecting the available photometry from literature .
bh masses have been recomputed using the h@xmath48 fwhm measurements reported in @xcite but differently from them , we used the de - absorbed x - ray luminosity @xcite instead of the continuum 5100 luminosity ( see discussion in sec .
[ sec : bh ] ) .
the latter can in fact be absorbed and/or dominated by the stellar continuum .
+ * _ 2 qsos hosted in ulirgs from @xcite _ + we include in our sample 2 x - ray obscured agn hosted in ulirgs observed with the keck laser guide star adaptive optics ( lgsao ) system and the osiris integral field spectrograph from @xcite .
the targets ( four in total ) were selected from optical and mir colors of the noao deep wide - field survey ( ndwfs ) of botes to be dust obscured galaxies ( dogs , * ? ? ? * ) and to have strong h@xmath5 emission .
bh masses have been recomputed using the formula derived in sec .
[ sec : bh ] and since only two sources are detected in the 5ks _ chandra _ x - ray observations of the botes field @xcite , we include only these two in our analysis .
as already pointed out by @xcite the use of x - ray luminosity leads to larger bh mass estimates , as expected by the fact that the rest - frame optical continuum is likely absorbed by dust .
stellar masses have been recomputed using the sed fitting technique as explained in sec .
[ sec : masses ] , after collecting the available photometry from literature . + * _ a qso from @xcite _ + we include in our sample also a qso at @xmath49 from @xcite .
this object has been selected as the most extreme x - ray - to - optical flux ratio ( exos ) sources amongst the sample of bright x - ray selected exos obtained by cross - correlating 2xmmp and sdss catalogs .
new ukirt optical / nir photometry and moircs ir spectroscopy have been presented in @xcite .
x - ray luminosity and fwhm of the broad h@xmath5 line , available in @xcite , have been used to compute the bh masses using the formula derived in sec .
[ sec : bh ] while stellar masses have been derived from the k - band absolute magnitude .
@xcite measures a k - band apparent magnitude of 19.9 in ab system , that converted in k - band absolute magnitude corresponds to m@xmath50=-25.74 .
assuming the relations between m@xmath10 and l@xmath50 for star - forming and quiescent galaxies found by @xcite , we estimated logm@[email protected]/l@xmath50 ratio of the sources of our sample for which the sed fitting procedure has been applied , fit well with this relation . ] . + lccc @xmath52 & @xmath53 & @xmath54 & @xmath21 + + cdfsx57@xmath55 & 03:32:12.9 & -27:52:36.9 & 2.561 + cdfsx153@xmath55 & 03:32:18.3 & -27:50:55.3 & 1.536 + cdfsx531@xmath55 & 03:32:14.4 & -27:51:10.9 & 1.544 + cdfsx633@xmath55 & 03:31:50.4 & -27:52:12.2 & 1.374 + + cosmos18@xmath56 & 10:00:31.9 & + 02:18:11.8 & 1.606 + cosmos60053@xmath56 & 10:01:09.2 & + 02:22:54.7 & 1.582 + cosmos175@xmath56 & 09:58:53.0 & + 02:20:56.4 & 1.530 +
cosmos2028@xmath56 & 10:02:11.3 & + 01:37:06.6 & 1.592 + cosmos5321@xmath56 & 10:03:08.8 & + 02:09:03.5 & 1.470 + + xmmes1_460@xmath57 & 00:36:41.5 & -43:20:38.1 & 1.748 + ab2690_29 @xmath58 & 00:01:11.4 & -25:12:05.1 & 2.087 + bpm1627_181@xmath58 & 00:50:31.6 & -52:06:30.0 & 1.335 + + a1236 + 6214 & 12:36:35.5 & + 62:14:24.0 & 2.015 + a1237 + 6203 & 12:37:16.0 & + 62:03:23.0 & 2.053 + a1312 + 4239 & 13:12:15.2 & + 42:39:00.0 & 2.555 + a1312 + 4238 & 13:12:22.3 & + 42:38:14.0 & 2.560 + a1636 + 4059 & 16:36:55.8 & + 40:59:14.0 & 2.592 + a1637 + 4053 & 16:37:06.5 & + 40:53:13.0 & 2.375 + + dog2 & 14:33:35.6 & + 35:42:43.0 & 1.300 + dog3 & 14:34:00.3 & + 33:57:14.0 & 1.684 + + 2xmmj1232 + 2152 & 12:32:04.91 & + 21:52:55.4 & 1.870 + as a comparison sample , we used a sample of @xmath1190 unobscured , optically type1 , blue qsos ( @xmath592 - 3 , * ? ? ?
* ) from the zcosmos survey in the redshift range [email protected] , presented in @xcite .
for consistency with our x - ray selection , only sources with xmm detections ( 82/89 ) have been included in the analysis .
@xcite computed their bh masses using the virial formula from @xcite . however , since we want to use these objects as comparison sample of the obscured one , we recomputed the bh masses using the mgii line and the 3000 continuum luminosity as described in sec [ sec : bh ] , by adopting the same calibration method as for the obscured sample ( eq . [ eq:2b ] ) . for blue , unobscured qsos
, the continuum l@xmath60 is not expected to be severely affected by obscuration and can therefore be used in the virial equation . for the stellar masses ,
since the method used by @xcite is the same used in this paper , we do not recompute them but we use the ones provided in their table 1 .
bh masses have been obtained using the relation which combines the virial theorem with the radius - luminosity relation and provides mass estimates which are dubbed as `` single epoch '' , in contrast to the mass estimates from reverberation mapping observations ( e.g. * ? ? ?
* ; * ? ? ?
the relation for single epoch mass estimates can be written as : @xmath61 where l@xmath62 is the luminosity estimator in units of 10@xmath63 erg s@xmath24 and _ v@xmath64 _ is the velocity estimator in units of 1000 km s@xmath24 .
@xmath15 is the slope of the radius - luminosity relation which connects the size of the broad line region ( blr ) with the continuum luminosity ( e.g. * ? ? ?
* ; * ? ? ?
the normalization @xmath5 is calibrated by assuming that the single epoch masses are in agreement with reverberation mapping masses @xcite which , in turn , are in agreement with the @xmath65 relation for normal galaxies @xcite .
such approach is motivated by the unknown geometry and kinematics of the blr ( see , e.g. , mclure & dunlop 2001 , onken et al .
however , due to the different choices adopted in the calibration , a large number of different relations are reported in the literature that relate black hole mass , broad line width and continuum luminosity ( e.g. @xcite ) .
the key point for this paper is that the bh - galaxy scaling relation adopted to calibrate @xmath15 should be the same or be consistent with the scaling relation at z=0 used as a reference for studying the redshift evolution .
here we adopt a relation which has been calibrated by directly imposing that single epoch masses agree with the local bh - galaxy scaling relations ( marconi et al .
2014 , in preparation ) .
briefly , single epoch virial products ( i.e. bh masses for @xmath66 ) have been estimated from h@xmath15 using the average spectra for the objects of the reverberation mapping database by @xcite updated with the measurements from the most recent campaigns . in the case of mgii
we have used the spectral measurements from international ultraviolet explorer ( iue ) data performed by @xcite .
bh masses have been estimated from the stellar velocity dispersion ( see e.g. * ? ? ? * for a compilation of data ) , or from the luminosity @xcite of the host spheroid using the @xmath65 and/or @xmath67 relations by @xcite .
the adopted relations can be expressed in the form @xmath68 @xmath69 where fwhm refers to the h@xmath15 and mgii broad lines and l@xmath70(5100 ) and l@xmath70(3000 ) represent the agn continuum luminosity @xmath71 at 5100 and 3000 , respectively .
uncertainties on the normalization , i.e. @xmath5 , are of the order of 0.1 dex ( see marconi et al .
2014 , in preparation , for more details ) . in the following , for the comparison with the @xmath72 relation , we will use as a reference the @xmath73 relation by @xcite because the sample adopted by those authors defines a @xmath74 relation consistent with that of @xcite . to compute the bh masses of the objects in our sample , we used eq .
[ eq:2 ] after converting h@xmath15 in h@xmath5 using the correlation between h@xmath5 and h@xmath15 widths found by @xcite .
moreover , since our sample is made of dust reddened qsos , the continuum luminosity at 5100 is expected to be absorbed and/or strongly contaminated by stellar continuum .
for this reason we replaced @xmath75(5100 ) with the absorption corrected x - ray @xmath76kev$ ] luminosity using eq .
( 5 ) of @xcite .
we obtain for the bh mass the following expression : @xmath77}}{10^{44 } erg \ , s^{-1 } } \label{eq:3}\ ] ] for our statistical analysis of the scaling relation evolution , we assign to each black hole mass measurement an error given by the sum of the statistical and systematic uncertainties .
the systematic uncertainty in the logm@xmath7 determination has been estimated in 0.3 dex to account for the observed scatter in the virial relation itself , while in the computation of the statistical errors , we take into account the errors in the x - ray luminosity and fwhm measurements , together with the uncertainties associated with the @xmath78}-l_{5100}$ ] relation coefficient which contributes with 0.07 dex ( in quadrature ) .
we measured the fwhm in the 9 new ir spectra presented here plus the three sources taken from @xcite for which we re - analyzed the spectra after improving the sky subtraction and extraction . following @xcite ,
the broad h@xmath5 component detected in the ir spectra has been fitted with a broad gaussian and one or more narrow gaussians accounting for the contribution of the narrow components of both h@xmath5 and [ nii ] .
the underlying continuum has been fitted with a linear component .
regions affected by low s / n and/or bad sky residuals are not taken into account in the fit .
we linked the narrow components of h@xmath5 and [ nii ] to have the same velocity shift and the same velocity dispersion .
moreover , the flux of the [ nii]@xmath796548 line was linked to be one third of the [ nii]@xmath796584 line .
the uncertainty in the width of the broad line is estimated by accounting for the correlation with the other components .
a zoom of the 12 analyzed spectra around the region of the broad h@xmath5 line , together with the line fit is shown in fig .
[ fig : spectra_fit ] . for the rest of the sources
, we have taken the fwhm published values .
rest frame fwhm values are reported in table [ tab : parameters ] . as explained above
, the measured fwhm of the h@xmath5 broad component has been used , together with the de - absorbed x - ray @xmath76kev$ ] luminosity in eq .
[ eq:3 ] to compute the bh masses . intrinsic x - ray [ 2 - 10kev ] luminosities have been computed after correcting the flux for the absorption n@xmath80 .
when possible , n@xmath80 values have been derived directly by fitting the x - ray spectra . on the contrary , for sources detected with few counts , for which a proper x - ray spectral analysis can not be performed ,
we have used the hardness ratio ( computed as hr=(h - s)/(h+s ) where s and h are the soft and hard band counts ) as a measure of obscuration . for the 9 sources for which new observations are presented , we have performed a fit of the x - ray spectra ( for details
see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) for all but cosmos60053 , for which the quality of the x - ray data is low . for these sources ,
the n@xmath80 is typically constrained within 0.1 dex .
this translates into uncertainties on the rest - frame x - ray fluxes ( and therefore luminosities ) from @xmath175% for the brightest sources to @xmath1120% for the faintest ones . for the remaining sources ,
the n@xmath80 values have been taken from literature or derived from the hardness ratio ( as specified in table [ tab : parameters ] ) assuming a power - law x - ray spectrum with a typical photon index @xmath81 = 1.8 @xcite .
since x - ray luminosities are used , together with the fwhm , to compute the bh masses , it is important to be sure that the derived values are robust .
to this end , we have used the mir / x - ray luminosity ratios @xcite to check the consistency of the l@xmath82 with the whole sed of the source ( see sec . [
sec : masses ] ) .
we find that the agreement is good for all but 6 sources . in three of these objects ,
the discrepancy is however relatively small ( corresponding to a @xmath110.3 dex difference in luminosity ) and can be ascribed to the large scatter present in the mir / x - ray ratio used @xcite . on the contrary , for the remaining three objects ( cdfsx633 , a1236 + 6214 and a1237 + 6203 ) , the discrepancy is larger i.e. @xmath110.5 @xmath83 0.6 dex in x - ray luminosity .
in these cases , there is the possibility that the x - ray luminosity as derived by the x - ray data may be not reliable due to an underestimate of the n@xmath80 value caused by the low counts statistics . for these sources , we derived new intrinsic sed - derived x - ray luminosities and we used them in the computation of the bh masses ( see next section ) .
intrinsic x - ray luminosities are reported in table [ tab : parameters ] . [ cols="<,^,^,^,^,^",options="header " , ]
in this section , we study and compare the rate at which host galaxies accrete in relation with their bhs . to this end
, we have derived the galaxy star formation rate and bh eddington ratio for each object in the sample . in general ,
the most reliable way to compute the galaxy s sfr is adding up the rate of unobscured star formation ( emitted in the uv ) with the obscured one re - emitted in the ( mid- and far- ) ir by dust ( sfr@xmath84 ) .
however , for bright ir sources , sfr@xmath85 is negligible compared to the absorbed one and we can simply assume @xmath86 .
for this reason , mid and far - ir measurements of cold dust heated by star - forming processes are in principle the most reliable way to measure the galaxy s sfr . on the one hand , in case of bright qsos , the agn contamination in the ir band can be not negligible and thus the derived sfrs slightly overestimated ( see e.g. * ? ? ?
* ; * ? ? ?
* ) . on the other hand ,
the optical / ir sed fitting procedure relies only on the uv emission measurements , which traces the unobscured sf and takes into account the obscured component using the dust correction factor computed from the full sed . for the objects in our sample we estimated the host star formation rates using submm / ir data when available and the sed fitting procedure otherwise .
the sfr of our sources are inferred from their infrared luminosities using the best set of mid / far - infrared / submm data available . for sources with pacs detections ,
ir luminosities are derived by fitting their far - infrared flux densities with the sed templates of @xcite . for sources with no far - infrared detection but with a submillimeter detection ( i.e. , from scuba ) , infrared luminosities
are derived from their scuba flux densities using the l@xmath87-submm relation of @xcite . from these ir luminosities , we infer sfrs assuming sfr @xmath88=10^{-10}\times l_{\rm ir}\,[l_{\odot}]$ ] @xcite for a @xcite imf .
= ssfr / ssfr@xmath89 of the host galaxies versus the bh eddington ratio .
the ssfr values are normalized assuming the best - fit of the galaxy main sequence as a function of redshift ( horizontal dashed line ) obtained by @xcite . dotted lines trace 4 times above and below the main sequence : the three delimited regions are used to define a galaxy to be starburst , main sequence or quiescent @xcite.,scaledwidth=45.0% ] sfrs have been derived from ir / or submm fluxes when available and from sed - fitting otherwise ( see table [ tab : sfr_edd ] ) .
open circles correspond to the cases for which we have only an upper limit on the sfr while filled circle with an arrow correspond to the objects with log(sfr ) @xmath170 .
the cross with the errorbar in the upper left corner shows the average error on the @xmath90-axis which corresponds to an error of @xmath110.35 dex in @xmath79 .
[ fig : scaling_z_sfr_edd ] for quantifying the bh accretion , we derived the eddington ratio , defined as the ratio between the agn bolometric luminosity and the eddington luminosity .
: @xmath91}{1.3 \times 10^{38 } \times m_{\rm bh}/m_{\odot}}\ ] ] where l@xmath92 have been derived from the x - ray luminosities using the bolometric correction derived from @xcite .
galaxy s sfrs and bhs eddington ratios @xmath79 are reported in table [ tab : sfr_edd ] .
figure [ fig : scaling_z_sfr_edd ] shows the starburstiness r@xmath93 = ssfr / ssfr@xmath89 ( ssfr@xmath89 being the reference ssfr = sfr / m@xmath10 of the galaxies in the main sequence ) of the host galaxies versus the bh eddington ratio .
open circles correspond to the cases for which we have only an upper limit on the sfr while filled circles with an arrow to the objects with log(sfr)@xmath170 . the starburstiness r@xmath93 @xcite measures the excess or deficiency in ssfr of a galaxy in terms of its distance from the galaxy main sequence ( ssfr@xmath89 ) . for this comparison
, we used the best - fit of the galaxy main sequence as a function of redshift obtained by @xcite .
following @xcite , we define a galaxy to be starburst if its sfr is above 4 times the main sequence ( r@xmath944 ) , quiescent if it is below 4 times the main sequence ( r@xmath951/4 ) and main sequence if 1/4@xmath964 ( see also * ? ? ?
* ; * ? ? ?
. moreover , we divided the sample in high accreting bhs ( @xmath970.2 ) and low accreting bhs ( @xmath980.2 ; see fig .
[ fig : scaling_z_sfr_edd ] ) .
we find that most ( 90% ) of our sources are hosted in starburst or main sequence star - forming galaxies with bhs accreting at @xmath970.1 ( 80% ) .
no clear trend is visible between bh accretion and host galaxy s sfr .
however , we note that starburst galaxies are preferentially found ( all but one ) to host high - eddington ( @xmath970.2 ) agn .
and m@xmath8 for obscured agn at [email protected] - 2.6 color - coded as explained in the inset of the plot .
the solid line is the local relation from @xcite .
for the 3 objects ( cdfsx633 , a1236 + 6214 and a1237 + 6203 ) for which the x - ray luminosity does not agree with the whole sed shape , we show two bh masses : color - coded symbols correspond to the correct sed - based l@xmath82 while the open symbols ( same symbols as the original ones ) are the bh masses computed assuming the ( lower ) l@xmath82 as measured from x - ray data . , scaledwidth=48.0% ] in figure [ fig : scaling ] , we show the m@xmath7 - m@xmath8 plane for our sample which spans the [email protected] - 2.6 range .
as explained in the previous section , for 3 sources ( cdfsx633 , a1236 + 6214 and a1237 + 6203 ) , the intrinsic x - ray luminosity disagrees with the rest of the sed .
these sources are suspected to be highly obscured ( possibly compton thick ) agn for which the measured n@xmath80 is underestimated due to the poor quality of the x - ray data and therefore the intrinsic luminosity much higher than that derived from the x - ray data .
since the intrinsic x - ray luminosity is used to compute the bh mass , the bh masses associated to these objects could be as well underestimated .
for these sources , we thus used the sed - derived l@xmath82 to derive the bh masses .
the values for the bh masses considering the l@xmath82 obtained from the x - ray data are shown in the plot with the same ( but open ) symbols as the corresponding subsample . since the virial relations used in this work to derive the bh masses are normalized to the @xcite local relation ( see sec .
3 ) , to study the redshift evolution , high redshift sources in this plane have to be compared to the above local relation .
the @xcite local relation is shown in fig .
[ fig : scaling ] with a solid line . from this comparison ,
we find that , x - ray obscured , red qsos are largely scattered around the local scaling relation . in particular , we note that while less massive objects are equally scattered below and above the local relation , the most massive ones are mainly found above .
considering the eddington ratios of the analyzed sources , we find both high and low accreting smbhs equally distributed below and above the local relation .
this is in contrast to the work from @xcite , who studied a sample of 13 luminous , red qsos at z@xmath171 , and found that low accreting smbhs ( @xmath980.3 ) lie mainly above the scaling relation while high accreting smbhs ( @xmath970.3 ) below .
moreover , we note that the position of some sources ( i.e. objects taken from @xcite and @xcite ) is different compared to what presented in their original papers .
this is due to a more accurate estimate of the stellar masses , a different calibration used to compute the bh masses , the use of the l@xmath82 instead of the l@xmath99 in the bh mass formula for the sources in @xcite and , for bpm1627_181 , also to a more accurate fwhm measurement .
differences in the bh mass are below 0.4 dex while the host masses can differ up to 0.7 dex . in fig .
[ fig : scaling_z ] , we plot the offset @xmath81(z ) between the m@xmath7/m@xmath8 and the local values , calculated as distance in the log space of each point to the @xcite correlation as the offset from the scaling relations in terms of `` excess black hole mass '' . to compare this measures with our points
we have divided these values by @xmath101 where a=0.79 is the slope of the @xcite relation ( see also * ? ? ?
] as a function of redshift : @xmath102 in this plot , we also include the comparison sample of optically type1 , blue qsos at [email protected] from @xcite for which we re - computed the bh masses using eq .
[ eq:3 ] of sec .
[ sec : bh ] , consistently with the dust obscured , red ones .
we see that both blue and red qsos show m@xmath7/m@xmath8 ratios increasing going to earlier epochs with red qsos having slightly lower values at [email protected] .
this is evident in the upper panel of fig .
[ fig : scaling_z ] where mean values and errors are shown for both populations . since the m@xmath10 of a1637 + 4053
is poorly constrained by the sed fit and its m@xmath7 is an upper limit due to its l@xmath82 , for the last bin of the red qsos sample , we show both the mean value taking it into account ( open circle ) and excluding it ( filled circle ) .
the solid and dashed lines correspond to the redshift evolution derived for blue qsos at z@xmath103 and extrapolated at higher redshift @xcite .
high redshift red qsos do not seem to follow such trend having on average lower values that remain constant at [email protected] .
the two distributions in the common redshift range ( [email protected] ) are shown in the right panel of fig .
[ fig : scaling_z ] . according to a kolmogorov - smirnov statistics ,
the probability that the two samples are drawn from the same parent population is 10% .
as discussed by e.g. @xcite and @xcite , agn flux limited samples are generally biased towards higher values of m@xmath7/m@xmath8 .
the magnitude of this bias depends on the flux limit of the sample i.e. it increases for brighter limiting fluxes , the underlying distribution functions and the intrinsic scatter in the m@xmath7-m@xmath8 relation . since both the unobscured agn sample from @xcite and the new sample of obscured agn presented here
are flux limited , i.e. x - ray flux limited from xmm - cosmos and cdfs , both samples are affected by this selection bias . however , since the chandra deep fields observations are very deep , the expected bias is negligible , contrary to the xmm - cosmos agn samples for which a non - negligible bias is expected ( as already discussed in * ? ? ?
we quantified the magnitude of the bias using the framework presented in @xcite .
we used their equation 29 , assuming the z@xmath1041 bh mass function and eddington ratio distribution function derived from the vvds survey ( schulze et al . in prep ) and applying the limiting fluxes corresponding to the considered fields . the result is shown in fig .
[ fig : scaling_bias ] . in this figure
, we show with grey points the averages and their errors of the offsets from the local relation of the unobscured type1 agn sample and with orange squares and the blue pentagon the averages for cdfs - only and cosmos - only obscured sources in the analyzed sample , respectively . while the average values with their errors for the cdfs - only and the cosmos - only sources in our sample are shown as orange squares and the blue pentagon , respectively .
note that the orange square at [email protected] is made of only one object .
the lines correspond to the expected luminosity bias computed considering the x - ray limiting fluxes of the x - shooter sample ( @xmath105 ; blue dot - dashed line ) , the sinfoni sample ( @xmath106 ; orange dashed lines ) and the xmm - cosmos sample of optically blue qsos ( long - dashed grey line).,scaledwidth=45.0% ] the predicted bias in case of no evolution is shown with a long - dashed grey line for the xmm - cosmos flux limit .
this has to be compared to the grey symbols . for the xshooter and sinfoni samples we consider as flux limits @xmath107 and @xmath108 respectively since above this fluxes the detection rate of broad h@xmath5 is @xmath1280% and in this respect the incompleteness of the sample is negligible ( see sec . [
sec : sample_new ] ) .
the predicted biases in case of no evolution for such limits are shown as a dot - dashed blue line in the first case and a dashed orange line for the latter case .
the dot - dashed blue line has to be compared to the blue point while the dashed orange line to the orange squares .
a slightly different treatment is needed for smgs and ulirgs which are primarily selected based on galaxy flux . for both samples ,
the additional agn selection criterion , based on the h@xmath5 flux , leads to a positive bias of the order of @xmath110.1 dex . as visible in the plot , the general trend for the sources of our sample as well as for the optical type1 agn from merloni
are reproduced by the bias predictions and would lead to the conclusion that both obscured and unobscured agn lie on the local scaling relation up to [email protected] . however , the observed samples show a higher mean value of @xmath109 ( in the case of cosmos ) due to observational biases , as described in @xcite .
interestingly , the cdfs - only mean value at [email protected] is much lower than the predicted one given the observational bias .
in fact , it is important to keep in mind that the estimates of bias effects can suffer from some limitations and uncertainties : ( i ) an accurate prediction of the sample bias at a given redshift requires the knowledge of the underlying distribution functions i.e. the spheroid mass function , the active bh mass function ( bhmf ) and the eddington ratio distribution function at that given redshift .
unfortunately , these distributions are still very poorly constrained at high @xmath21 , i.e. the spheroid mass function itself is basically unknown at high @xmath21 . also the bhmf is best established at the high mass end while the low mass end is less well determined and the systematics not fully understood . moreover , ( ii ) even considering all these uncertainties , a reliable bias prediction requires a sample with a well defined selection function .
this is the case for the optical type1 , blue agn sample from merloni which are selected from the whole xmm - cosmos sample to be at [email protected] and with broad lines in the spectra .
the same is not true for our sources , made of different subsamples of objects for which spectroscopic nir follow - up was performed .
in this section , we compare our observational results with the semi - analytic model from ( * ? ? ? * and reference therein ) which allows us to follow the accretion onto smbhs and the related agn activity together with the evolution of galaxies . according to this model , galaxy interactions ( i.e. both major and minor merging and fly - by events ) trigger both starburst events and smbh growth by inflow of cold gas in the galaxy disk which is destabilized and thus loses angular momentum .
in addition , a second component of quiescent star formation , corresponding to the gradual conversion of gas into stars on a longer timescale ( @xmath110 compared to @xmath1110@xmath111yr of the starburst events ) , is always ongoing and responsible for the galaxy growth .
the amount of gas available in the system , together with the galaxy interaction rate and their effectiveness in destabilizing the gas , are hence the driving parameters regulating both galaxy and smbh growths .
this model correctly describe the tight correlation between the bh mass and the stellar mass in the local universe @xcite and the evolution of the luminous agn population over a wide range of redshift ( see * ? ? ?
* ; * ? ? ?
such observables are also reproduced by several other semi - analytic models based on the assumption of interaction - triggered agn implemented with accretion of hot gas and/or of instabilities in self - gravitating disks as additional triggers ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we compare our observations with the menci model since it is ideally suited to our purpose .
it is indeed the only one that includes a physical description of agn feedback providing a distribution of the agn absorbing column density as a function of the luminosity and redshift tested against observations ( see * ? ? ?
* ) . by following the merging history of the galaxies at different epochs ,
it is possible to study the relative growth of bhs and host galaxies and to make predictions on the m@xmath7-m@xmath112 relations at different epochs and as a function of various galaxy properties e.g. mass , gas fraction and star formation rate .
@xcite showed that a typical prediction of such interaction - driven model is that the evolution of the m@xmath7-m@xmath8 relation is stronger for increasing stellar or bh mass .
moreover , in agreement with the observations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , they find that blue qsos have a higher m@xmath7/m@xmath8 ratio at earlier epochs which decreases going to lower redshift and approaches the local relation .
this trend is due to the fact that , in the early phases , the assembly of bh masses is extremely rapid and much faster than the stellar mass growth since these objects are hosted in the most massive halos formed in dense environments in which both the interaction rate and the fraction of destabilized gas are large . on the contrary , going towards lower redshift , the decline of the interaction rate and of the destabilized gas fraction , suppresses the growth of bhs which is only due to galaxy interactions while quiescent star formation still proceeds continuing to build up stellar mass .
( z ) of n@xmath110@xmath3 @xmath4 and @xmath29 agn ( ( * ? ? ? * and reference therein ) ; see also @xcite ) .
the three filled contours correspond to different values of the fraction of objects with a given value of @xmath81(z ) at the considered redshift : 0.02 , 0.1 and 0.2 from the lightest region to the darkest .
the colored dots with error bars denote our selected sample with the same coding as in fig .
[ fig : scaling].,scaledwidth=45.0% ] in this section , we take a step forward and we follow in the monte carlo simulations only the x - ray obscured , reddened qso .
the model includes a detailed treatment of agn feedback which is directly related to the impulsive , luminous agn phase @xcite and is based on expanding blast waves as a mechanism to propagate outwards the agn energy injected into the interstellar medium at the center of galaxies . in the simulation ,
agn are classified as obscured ( n@xmath80 @xmath113 10@xmath3 @xmath4 ) and unobscured ( n@xmath11410@xmath3 @xmath4 ) on the basis of the neutral hydrogen column densities corresponding to the unshocked gas in the galaxy disk ( equation 6 in * ? ? ?
* ) . since our sample is not a complete sample of sources with a clean selection but rather a collection of sources selected with slightly different criteria , it is not easy to compare them with the model predictions .
x - ray obscuration and dust - reddening are common properties of all sources . for this reason , to select in the simulation the sample which is most similar to the data presented in this work , we performed a color cut selecting only x - ray obscured sources with red colors ( i.e. r@xmath83k@xmath124 in vega magnitudes ) .
the simulated galaxy magnitudes are computed from the predicted star formation and chemical enrichment histories using the single stellar population model of @xcite and the salpeter initial mass function .
the dust extinction affecting the above magnitudes is computed from the dust optical depth and applying the appropriate attenuation to the luminosity at various wavelengths .
the comparison is shown in fig .
[ fig : scaling_z_model ] where , together with the observational points ( symbols and color - code are as in fig .
[ fig : scaling ] . ) , we plot the contour plot of the predicted redshift evolution of the offset from the local m@xmath7@xmath83m@xmath8 relation ( @xmath81(z ) ) for agn with log([email protected] , n@xmath110@xmath3 @xmath4 and @xmath29 .
these cuts are chosen to best reproduce the selection criteria of the observed sample .
the three contour levels correspond to different values of the fraction of objects with a given value of @xmath81(z ) at the considered redshift : 0.02 , 0.1 and 0.2 from the lightest region to the darkest .
the model predicts that the @xmath109 value for obscured and red qsos slightly increases towards higher m@xmath7/m@xmath8 , with a lower offset than the one predicted for blue qsos ( see * ? ? ?
this is in excellent agreement with our observations .
we do actually find that most of the sources in our sample lie well inside the contours of the model prediction and that red x - ray obscured agn and optical type1 , blue agn follow a very similar path in the m@xmath7-m@xmath8 plane .
in this work we have analyzed a sample of x - ray obscured , intermediate type , dust reddened qsos at [email protected] .
red , obscured qsos are different from optical type1 blue qsos studied so far both in their host and nuclear properties , in that they are dusty and thus red and their nuclei are obscured .
these objects may represent a `` transitional '' phase in the agn - galaxy co - evolving scenario as expected by the popular theoretical models by e.g. @xcite . according to these models , in fact , major mergers are the main agn fueling mechanism and galaxy mergers are able to funnel a large amount of gas onto the nuclear regions in a short time scale . in this phase the agn
is indeed expected to be obscured and red .
moreover , the resulting high gas density in the central region of the galaxy triggers starburst events : red and obscured qsos are expected to live in star - forming galaxies as found for most of our sources ( 90% are main - sequence or starburst galaxies ) .
later , as radiative pressure on the dust grains begins to clear the dust and gas away during the blowout phase in which evidence of outflows are expected , the agn is revealed as an uv - luminous optical type1 , blue qsos .
we have analyzed the position of these sources in the @xmath100 plane .
we found a trend with the bh mass , i.e. while less massive red qsos are about equally scattered below and above the local relation , the most massive ones are mainly located above it . the same trend with bh mass
is predicted by the interaction - driven models since high mass bhs form in the most biased regions of the primordial density field where high redshift interactions were favored ( see e. g. * ? ? ?
looking at the average m@xmath7/m@xmath10 ratios as a function of redshift , we find that , similarly to unobscured , optically blue qsos ( e.g. merloni et al . , 2010 ) , obscured red qsos show an higher ratio compared to the local one , and the increase is higher going to higher redshift ( [email protected] ) .
possible observational biases have been analyzed .
in particular , as discussed by e.g. @xcite and @xcite , agn flux limited samples are generally biased towards higher values of m@xmath7/m@xmath8 .
looking at the xmm - cosmos samples and the cdfs sample , we find that the observed general trend in the @xmath116 plane is very similar to the one predicted by the bias , thus suggesting that both obscured and unobscured agn lie on the local scaling relation up to high z but are observed at higher values of @xmath109 due to observational biases .
+ however , we must keep in mind that the estimates of bias effects suffer from some limitations i.e. they require the knowledge of the underlying distribution functions ( the spheroid mass function , the active bh mass function and the eddington ratio distribution function ) at the given redshift and unfortunately , these distributions are still very poorly constrained at high @xmath21 .
the computed @xmath117 average values are consistent with or slightly lower ( up to [email protected] ) compared to what has been found for blue qsos @xcite .
this result suggests that in the analyzed population of obscured , red qsos the bh and their host stellar masses are already fully formed and thus their ratio is similar to what observed in optically blue qsos .
recently @xcite studying a sample of 13 z@xmath171 luminous , dust reddened qsos , found that these sources are preferentially located below the local scaling relation .
this result has been interpreted as the evidence that these sources are in the intermediate phase ( blow - out phase , see e.g. @xcite ) between the major - merger induced starbursts which appear as ulirgs and smgs and the optical type1 , blue qsos where we expect a rapid bh growth . in our sample , on the contrary , the bh has already grown up and the objects are located on average above the local relation , especially at the high mass end .
however , strong outflows are still visible i.e. for the sources for which we have high resolution spectra with good s / n of the [ oiii ] region from x - shooter , we indeed find compelling evidence of the presence of outflowing material in the ionized gas component ( brusa et al . , 2014 ) .
this suggests that we are observing the final stage of such intermediate phase , when the bh growth is at its end but the agn feedback has still not finished to clean up the dust and nuclear gas .
this is not so surprising considering that these sources do show broad lines in their optical lines ( redshifted in the nir ) : at least part of the obscuring material has already cleared - up and the agn broad line regions are visible .
larger samples expanding also to higher redshift are necessary in order to consolidate such results .
large area x - ray surveys with associated multiwavelength follow - up , such as xxl @xcite and stripe-82 @xcite can be exploited to this end . moreover , ifu spectroscopy and submm observations are fundamental to study in detail such rare objects and to provide a direct observational proof of quasar feedback through the detection of possible galactic scale outflow and on its effect on the host galaxy star formation by mapping the spatial distribution of the sfr ( see e.g. * ? ? ?
we thank the referee for the careful reading and the extremely useful comments provided which help to strenght the presented results .
this work is based on observations made at the european southern observatory , paranal , chile ( eso program 88.b-0316(a ) and 090.a-0830(a ) ) .
ab work is supported by the inaf - fellowship program .
mb acknowledge support from the fp7 career integration grant `` eeasy '' ( cig 321913 ) .
support for this publication was provided by the italian national institute for astrophysics ( inaf ) through prin - inaf 2011 `` black hole growth and agn feedback through the cosmic time '' and by the italian ministry for school , university and research ( miur ) through prin- miur 2010 - 2011 `` the dark universe and the cosmic evolution of baryons : from current surveys to euclid '' .
mcd work was supported by the marie curie training network elixir under the contract pitn - ga-2008 - 214227 from european commission and is currently supported by conacyt ( ciencia bsica ) grant 167332-f and by papiit - unam grant ia100212 . | we present near - infrared spectra , obtained with ` sinfoni ` and ` xshooter ` observations at eso vlt , of nine dusty , red qsos at @xmath0 .
the sources are hard x - ray detected , characterized by cold absorption ( n@xmath110@xmath2 - 10@xmath3 @xmath4 ) and show a broad h@xmath5 component in the nir spectra .
we complement this sample with twelve additional sources taken from the literature with similar properties resulting in a total sample of 21 x - ray obscured , intermediate type ( 1.8 - 1.9 ) , dusty reddened qsos . from the broad h@xmath5 line we have computed the bh masses through the virial formula and derived eddington ratios .
moreover , from optical / ir multi - component sed fitting we have derived the stellar mass of their host galaxies and their sfrs .
we find that most of the sources in our sample are hosted in starburst and main sequence star - forming galaxies with eddington ratios @xmath6 .
we find a strong trend with the bh mass i.e. less massive objects are scattered below and above the local relation while the most massive ones are mainly located above it .
we also studied the evolution of these sources on the m@xmath7-m@xmath8 plane compared to a sample of optically blue type1 qsos and we find that obscured red qsos show a ratio of m@xmath7 to m@xmath8 that increases with redshift which is consistent with or slightly lower than what has been found for blue qsos .
these sources may represent the blow - out phase at the end of the rapid bh growth and immediately preceding the classical blue qsos typically sampled in optical surveys .
they in fact show evidence of outflows in the ionized gas component , but their bh has already fully formed .
[ firstpage ] galaxies : active - galaxies : evolution - quasars : emission lines - quasars : supermassive black holes - cosmology : observations |
an _ alternating sign matrix _ ( @xmath4 ) is a square matrix with entries @xmath3 , @xmath6 or @xmath5 such that along each row and each column the non - zero entries alternate and add up to @xmath6 .
an example is given next .
@xmath7 ( the coloring of the entries is explained shortly . )
the story of @xmath4s began in the 1980 s when mills , robbins and rumsey @xcite defined them in the course of generalizing the determinant and conjectured that the number of @xmath8 @xmath4s is given by the following simple product formula .
@xmath9 it was more than ten years later when zeilberger @xcite finally succeeded in providing the first proof of this formula in an @xmath10 page paper .
kuperberg @xcite then used six - vertex model techniques to give a shorter proof . as early as in the 1980 s , as discussed by robbins (
* ; * ? ? ?
* , p. 2 ) , stanley made the suggestion to systematically study _ symmetry classes of @xmath4s _ , which led robbins to numerous conjectures , see @xcite .
in particular , several symmetry classes of @xmath4s were conjectured to be enumerated by beautiful product formulas similar to .
the program of proving these product formulas was recently completed in @xcite , in which @xmath4s of odd order that are invariant under the reflections in the diagonal and in the antidiagonal , usually referred to as _ diagonally and antidiagonally symmetric @xmath4s _ ( @xmath1s ) , were enumerated .
the example in belongs to this symmetry class .
about half of robbins other conjectured product formulas for symmetry classes of @xmath4s were proven by kuperberg in @xcite , namely those for _ vertically symmetric @xmath4s _ , _ half - turn symmetric @xmath4s _ of even order and _ quarter - turn symmetric @xmath4s _ of even order .
razumov and stroganov proved the odd order cases for half - turn symmetric @xmath4s @xcite and for quarter - turn symmetric @xmath4s @xcite , and okada enumerated _ vertically and horizontally symmetric @xmath4s _ @xcite .
in addition to symmetry classes of @xmath4s , various closely - related classes of asms have also been studied .
of relevance for this paper are _ off - diagonally symmetric @xmath4s _ ( osasms ) , and _ off - diagonally and off - antidiagonally symmetric @xmath4s _ ( oosasms ) , as introduced by kuperberg @xcite .
osasms are even order diagonally symmetric asms in which each entry on the diagonal is @xmath3 , while oosasms are dasasms of order @xmath11 in which each entry on the diagonal and antidiagonal is @xmath3 .
a product formula for the number of osasms ( which is identical to that for odd order vertically symmetric asms ) was obtained by kuperberg @xcite , but no formula for the plain enumeration of oosasms is currently known . for further information regarding symmetry classes and related classes of asms ,
see , for example , ( * ? ? ?
1.21.3 ) , and references therein .
the purpose of the current paper is the study of odd order @xmath1s with a certain extreme behavior along the union of the diagonal and the antidiagonal .
observe that a @xmath1 @xmath12 of order @xmath13 is determined by its entries in the _ fundamental triangle _
@xmath14in the example marked with red .
for a given odd - order @xmath1 @xmath15 and @xmath16 , let @xmath17 where here and in the following `` diagonals '' refers to the union of the ( main ) diagonal and the ( main ) antidiagonal .
in the example , @xmath18 , @xmath19 and @xmath20 .
we have the following bounds for these statistics .
( the proof of the proposition is provided in subsection [ prop ] . )
[ bounds ] for any @xmath21 @xmath1 @xmath15 , the statistics @xmath22 lie in the following intervals .
1 . @xmath23 2 .
@xmath24 3 .
@xmath25 all inequalities are sharp .
the research presented in this paper started out with numerical data providing evidence that for four out of the six inequalities in the proposition we have the following phenomenon : the number of @xmath1s where equality is attained is round and in fact equal to numbers that have previously appeared in plane partition or alternating sign matrix counting .
it is the primary objective of this paper to prove all these empirical observations .
a number of generalizations including determinant or pfaffian formulas for certain generating functions that are known as _ partition functions _ in a physics context are also provided .
next we state the main results .
order @xmath13 @xmath1s @xmath15 with @xmath26 are proven to be equinumerous with @xmath8 @xmath4s .
it is remarkable that this now establishes a new class of objects with this property .
the other currently - known classes are _ totally symmetric self - complementary plane partitions _ in an @xmath27 box ( which were introduced by stanley @xcite and enumerated by andrews @xcite , and for which the equinumeracy with @xmath4s was first conjectured by mills , robbins and rumsey @xcite ) and _ descending plane partitions _ ( @xmath28 ) with parts no greater than @xmath29 ( which were introduced and enumerated by andrews @xcite , and for which equinumeracy with @xmath4s was first conjectured by mills , robbins and rumsey @xcite ) .
we are also able to identify statistics that have the same distribution : for an @xmath4 @xmath15 , let @xmath30 and , for an order @xmath13 @xmath1 @xmath15 , let @xmath31 throughout the paper , the sets of order @xmath29 @xmath4s and order @xmath29 @xmath1s are denoted by @xmath32 and @xmath33 , respectively .
[ maxmone ] the distribution of the statistic @xmath34 on the set @xmath32 is equal to the distribution of the statistic @xmath35 on the set of @xmath36 with @xmath26 , i.e. , for all non - negative integers @xmath37 we have @xmath38 a corresponding result with the set of @xmath36 such that @xmath26 replaced by the set of @xmath28s with parts no greater than @xmath29 and @xmath35 replaced by the number of special parts in the @xmath28was conjectured by mills , robbins and rumsey @xcite and proven by behrend , di francesco and zinn - justin @xcite .
in fact , they have proven a refinement ( also conjectured by mills , robbins and rumsey ) that involves two additional statistics . on the @xmath4 side ,
these are the _ inversion number _ and the _ position of the unique @xmath6 in the top row_. ( this was further generalized in @xcite , where they also included the position of the unique @xmath6 in the bottom row . ) in section [ ast ] , we define an inversion number @xmath39 on the set of @xmath40 with @xmath26 such that the joint distribution of @xmath34 and @xmath41 on @xmath32 is equal to the joint distribution of @xmath35 and @xmath39 on this subset of @xmath42 ( theorem [ maxmone - gen ] ) , thus establishing a refinement of theorem [ maxmone ]
. other generalizations of theorem [ maxmone ] in terms of partition functions for certain triangular six - vertex configurations are provided in theorems [ astpartfunct ] and [ astschurtheo ] , and in corollary [ astpartfunct1 ] . in the second theorem ,
_ cyclically symmetric plane partitions _ ( @xmath44 )
make an appearance .
they were enumerated by andrews @xcite .
[ maxone ] the number of @xmath36 with @xmath45 is equal to the number of @xmath44s in an @xmath46 box . the number of @xmath44s has previously appeared in the program of enumerating symmetry classes of @xmath4s : robbins conjectured @xcite and kuperberg ( * ? ? ?
* theorem 2 ) proved that the number of @xmath47 half - turn symmetric alternating sign matrices is the product of the total number of @xmath8 @xmath4s and the number of @xmath44s in an @xmath46 box .
generalizations of theorem [ maxone ] in terms of partitions functions are provided in theorems [ dastpartfunct ] and [ dastschurtheo ] , and in corollary [ dastpartfunct1 ] . as for the lower bound of @xmath49
, we again have the total number of @xmath4s turning up .
this is one of the few exceptional cases in this field that can easily be proven by establishing a bijection with other objects that are known to be enumerated by these numbers , in this particular case to the order @xmath50 objects of theorem [ maxmone ] , see subsection [ imply ] . as it is always the situation in such a case in this area so far
, the bijection is almost trivial , which is the reason why the following result can be viewed as a corollary of theorem [ maxmone ] .
[ minzero ] the number of @xmath36 with @xmath51 is equal to the number of @xmath4s of order @xmath50 . in the case of the upper bound of @xmath49 , the number of _ vertically and horizontally symmetric alternating sign matrices _
( @xmath53s ) appears .
[ maxzero ] the number of @xmath36 with @xmath54 is equal to the number of order @xmath55 @xmath53s .
the @xmath1s @xmath15 of order @xmath13 with @xmath54 can be regarded as odd order versions of @xmath56s , and so we denote this subset of @xmath42 by @xmath57 .
indeed , as the central entry of an odd order @xmath1 is always non - zero , all other entries on the diagonals of such an @xmath1 are zero and thus this is for odd order as close as one can get to kuperberg s original @xmath56s . of an @xmath58
is @xmath59 : the sum of entries in @xmath15 is certainly @xmath13 , however it is also @xmath60 , where @xmath61 is the sum of entries in the fundamental domain of @xmath15 without @xmath62 . ] interestingly , vertically symmetric @xmath56s of odd order have been enumerated by okada ( * ? ? ?
* ( b2 ) and ( b3 ) of theorem 1.3 ) .
( there they are referred to as @xmath63s , since the vertical symmetry and the diagonal symmetry implies the antidiagonal symmetry . )
generalizations of theorem [ maxzero ] in terms of partition functions are provided in theorems [ qoosasmpartfunct ] and [ spq ] . in section [ basics ] , we provide several basic observations : we prove proposition [ bounds ] , and , already for this purpose , it is useful to translate our problems into the counting of certain orientations of triangular regions of the square grid ( _ triangular six - vertex configurations _ ) . in this section
, we also present a simple bijection between the order @xmath29 objects of theorem [ maxmone ] and the order @xmath64 objects of corollary [ minzero ] that consists merely of manipulations close to the diagonal and the antidiagonal . in section [ weight - sec ]
, we introduce the vertex weights and some of their properties ( _ yang baxter equation _ , _ reflection equations _ ) , and use these weights to define the _ partition function _ ,
i.e. , a multiparameter generating function of the objects we want to count .
there we also introduce several specializations of this partition function that are used to prove our theorems .
the partition function is a laurent polynomial in the so - called _ spectral parameters _ , and , in section [ char - part ] , we provide characterizations of this partition function that are used later on . in sections [ ast ] [ qoosasm ] , we then employ all these preparations to prove theorems [ maxmone ] , [ maxone ] and [ maxzero ] , respectively .
finally , in appendix a we provide an ad - hoc counting of alternating sign triangles with one @xmath5 .
let @xmath65 be a @xmath1 of order @xmath13 .
its restriction to the fundamental triangle @xmath66 is said to be an _
odd @xmath1-triangle _ of order @xmath29 .
an example of an odd @xmath1-triangle of order @xmath67 is given next .
@xmath68 in fact , a triangular array @xmath66 of this form , in which each entry is @xmath3 , @xmath6 or @xmath5 , is an odd @xmath1-triangle of order @xmath29 if and only if , for each @xmath69 , the non - zero entries in the following sequence alternate read from top left to the top right and add up to @xmath6 .
@xmath70 ( in the example , this sequence is indicated in red for @xmath71 . )
let us clarify that in the special case @xmath72 , we require that the sequence @xmath73 has this property , and this is satisfied if and only if the non - zero entries of @xmath74 alternate , the first non - zero entry in this sequence is @xmath6 , and @xmath75 .
for an odd @xmath1-triangle @xmath15 , we define @xmath76 , where @xmath77 is the @xmath1 corresponding to @xmath15 . in the _ six - vertex model _ , these triangular arrays correspond to orientations of a triangular region of the square grid with @xmath78 centered rows as indicated in figure [ dasasmgraph ] , where * the degree @xmath79 vertices ( _ bulk vertices _ ) have two incoming and two outgoing edges , and * the top vertical edges point up . the term _ six - vertex _ is derived from the fact that there are @xmath80 possible local configurations around a bulk vertex .
the underlying undirected graph is denoted by @xmath81 in the following .
the triangular six - vertex configuration of an odd @xmath1-triangle can be obtained by restricting the standard six - vertex configuration on a square of the respective @xmath1 to the fundamental triangle . more concretely ,
each vertex of @xmath81 , except for the degree @xmath6 vertices at the top , corresponds to an entry of the associated odd @xmath1-triangle of order @xmath29 . as usual , a bulk vertex whose two incident vertical edges are outgoing ( @xmath82 ) , respectively incoming ( @xmath83 ) , corresponds to a @xmath6 , respectively @xmath5 , in the odd @xmath1-triangle , while all other bulk vertices correspond to zeros .
_ left boundary vertices _ and _ right boundary vertices _ are the degree @xmath84 vertices on the left and right boundary , and such a vertex corresponds to a @xmath6 ( resp .
@xmath5 ) if and only if the local configuration around the vertex is the restriction of the local bulk configuration corresponding to a @xmath6 ( resp .
@xmath5 ) .
this also applies to the bottom vertex .
that is , @xmath85 the configuration in figure [ dasasmgraph ] is the six - vertex configuration of the odd @xmath1-triangle in .
note that when restricting the six - vertex configuration of an odd @xmath1 to the fundamental triangle , the local configuration @xmath86 on the left - boundary must originate from @xmath82 , and thus corresponds to @xmath6 , because the only other local configuration with this restriction , i.e. , @xmath87 , can not appear on the diagonal of a six - vertex configuration of a @xmath1 .
we have a similar situation for @xmath88 . in order to derive the crucial identity
, we employ the fact that , in a directed graph , the sum of all outdegrees is equal to the sum of all indegrees .
the bulk vertices of a triangular six - vertex configuration as well as the vertices on the left and right boundary that correspond to @xmath89 s are balanced in the sense that the outdegree is equal to the indegree , and so these vertices do not contribute to this identity . for an odd @xmath1-triangle @xmath15 ,
we denote the number of boundary zeros with indegree @xmath84 in the corresponding triangular six - vertex configuration by @xmath90 , and number of boundary zeros with outdegree @xmath84 by @xmath91 . in the example in figure [ dasasmgraph ]
, we have @xmath92 and @xmath93 .
now , since all @xmath13 top vertices have indegree @xmath6 , the sum of all outdegrees is @xmath94 $ ] ( where we use the iverson bracket , i.e. , @xmath95=1 $ ] if the statement is true , and @xmath3 otherwise ) , while the sum of all indegrees is @xmath96 $ ] , and , since @xmath97-[a_{n+1,n+1}=-1]=a_{n+1,n+1}$ ] , we can conclude that @xmath98 the lower bounds for @xmath99 are trivial and the upper bound for @xmath100 follows because the central entry of a @xmath1 is always non - zero .
an example of a @xmath1 in which the upper bound of @xmath49 is attained is the matrix where every other entry of the restriction of the superdiagonal ( i.e. , the secondary diagonal immediately above the main diagonal ) to the fundamental triangle is @xmath6 , the central entry is @xmath59 and all other entries of the fundamental triangle are @xmath3 .
next we display the corresponding odd @xmath1-triangles of order @xmath79 and @xmath101 .
@xmath102 as we see below that @xmath26 implies @xmath103 , and @xmath43 implies @xmath104 , the fact that the two trivial inequalities are sharp follows from the fact that the upper bounds of @xmath105 are sharp .
the latter facts are shown below .
_ case @xmath106 : _ using @xmath107 and , we deduce @xmath108 we have equality iff @xmath109 and @xmath110 , or else @xmath92 and @xmath111 .
however , the latter possibility can not occur since then the bottom bulk vertex in the central column would have three incoming edges the left ( resp .
right ) boundary vertex adjacent to it is of type @xmath112 or @xmath113 ( resp .
@xmath114 or @xmath115)and this is impossible .
an order @xmath13 @xmath1 with @xmath26 is @xmath116 _ case @xmath117 : _ here we have @xmath118 and equality is attained iff @xmath92 , @xmath104 and @xmath119 . the identity matrix is a matrix where equality is attained .
_ case @xmath120 : _ from , it follows @xmath121 and so we have equality iff @xmath92 and @xmath119 . since @xmath43 implies @xmath48 , this inequality is sharp too .
the proof of proposition [ bounds ] implies immediately the following characterization .
[ char ] let @xmath15 be an odd @xmath1-triangle of order @xmath29 . 1
. [ char-1 ] we have @xmath26 if and only if , for each row except the bottom row , the sum of entries is @xmath6 when disregarding the left and right boundary entries . in the six - vertex configuration , this is fulfilled if and only if , in each row , the leftmost horizontal edge points to the right and the rightmost horizontal edge points to the left .
[ char-2 ] we have @xmath43 if and only if the sum of entries in each row is @xmath6 . in the six - vertex configuration , this is fulfilled if and only if , in each row , the leftmost vertical edge and the rightmost vertical edge point upwards .
[ char-3 ] we have @xmath48 if and only if , for each row , one of the following is true
the sum of entries is @xmath6 .
the sum of entries with the left or right boundary entry excluded from the sum is @xmath6 .
the sum of entries with both boundary entries excluded from the sum is @xmath6 . 1
. odd @xmath1-triangles @xmath15 of order @xmath29 with @xmath26 are precisely the odd @xmath1-triangles of order @xmath29 in which the sum of entries is minimal .
indeed , the characterization in corollary [ char ] ( [ char-1 ] ) implies the following different characterization : an odd @xmath1-triangle @xmath15 of order @xmath29 satisfies @xmath26 if and only if each column sum is @xmath3 . in order to see this , also recall that the bottom entry of such an @xmath15 is @xmath5 .
now the sum of entries in an odd @xmath1-triangle is at least @xmath3 ( since each column sum is at least @xmath3 ) , and the minimum is attained if and only if each column sum is @xmath3 .
the characterization in corollary [ char ] ( [ char-2 ] ) implies that odd @xmath1-triangles @xmath15 of order @xmath29 with @xmath43 are precisely the odd @xmath1-triangles of order @xmath29 in which the sum of entries is maximal .
this is because the sum of entries in an odd @xmath1-triangle of order @xmath29 is at most @xmath50 , since each of the @xmath50 rows has row sum at most @xmath6 , and the maximum is attained if and only if each row has sum @xmath6 .
this is best understood in terms of the six - vertex model . as noted in the proof of proposition [ bounds ] , the triangular six - vertex configurations equivalent to the objects from corollary [ minzero ] are characterized by the two facts that there is no left or right boundary vertex with indegree @xmath84 and the bottom vertical edge points upwards . to transform an order @xmath29 configuration of corollary [ minzero ] into an order @xmath50 configuration of theorem [ maxmone ] , add vertices left of each left boundary vertex and connect the new vertices to their right neighbors by an edge that is directed to the right , see figure [ implies ] for an example .
add similar vertices and edges on the right boundary , where the new horizontal edges are directed to the left . also add a new vertex below the bottom central vertex and introduce vertical edges that connect the @xmath55 added vertices to their top neighbors ( for the two new top boundary vertices add vertices above them and two vertical edges that are directed upwards ) . since the indegree of each former boundary vertex was either @xmath3 or @xmath6 ( and is currently either @xmath6 or @xmath84 ) and the former bottom vertex had indegree @xmath3 ( and has currently indegree @xmath84 ) ,
there is a unique way to orient the new vertical edges such that the former boundary and bottom vertices ( now bulk vertices ) are balanced . by corollary [ char ] , this produces an object with the desired properties . to reverse the transformation , simply delete left and right boundary vertices of an order @xmath50 configuration of theorem [ maxmone ] as well as the bottom vertex , and all edges incident with these vertices .
none of the new left or right boundary vertices can have indegree @xmath84 , since , before the deletion of the vertices and edges , the leftmost vertical edge in each row was pointing to the right , while the rightmost vertical edge in each row was pointing to the left .
this also implies that the bottom bulk vertex in the central column was of type @xmath82 before the deletion , and thus the new bottom vertex points upwards .
if we delete the diagonals of an odd @xmath1-triangle @xmath15 of order @xmath29 with @xmath26 , then , by corollary [ char ] , we obtain an object of the following type .
an _ alternating sign triangle _ ( @xmath122 ) of order @xmath29 is a triangular array @xmath123 in which each entry is @xmath3 , @xmath6 or @xmath5 and the following conditions are fulfilled . 1 . [ ast-1 ] the non - zero entries alternate in each row and each column . 2 . [ ast-2 ]
all row sums are @xmath6 .
[ ast-3 ] the topmost non - zero entry of each column is @xmath6 ( if it exists ) .
the set of order @xmath29 @xmath122s is denoted by @xmath124 .
here is a list of all @xmath122s of order @xmath125 .
@xmath126 in fact , order @xmath29 @xmath122s are in bijection with odd @xmath1-triangles of order @xmath29 with @xmath26 as the diagonals can be reconstructed as follows : @xmath127 the resulting triangle is surely an odd @xmath1-triangle @xmath77 of order @xmath29 , as the non - zero entries of all sequences as given in @xmath128 are alternating and add up to @xmath6 ; furthermore , it is the unique way of adding entries in these positions to achieve that .
moreover , we have @xmath129 as an @xmath122 has precisely @xmath29 columns that add up to @xmath6 .
this is because the total sum of all entries in an order @xmath29 @xmath122 is @xmath29 ( by property in the definition ) and each of the @xmath130 column sums is either @xmath3 or @xmath6 ( by properties and ) .
theorem [ maxmone ] states that , for all non - negative integers @xmath131 , the number of @xmath8 @xmath4s with @xmath132 occurrences of @xmath5 is equal to the number of @xmath122s of order @xmath29 with @xmath132 occurrences of @xmath5 . for @xmath133 ,
this is easy to see : @xmath4s without @xmath5 are permutation matrices .
there are also @xmath134 @xmath122s of order @xmath29 : each row contains precisely one @xmath6 and we build up the @xmath122 by placing in each row a @xmath6 , starting with the bottom row . for the @xmath6 in the bottom row , there is one choice , for the @xmath6 in the penultimate row there are in principal @xmath125 possible columns , but one is already taken by the @xmath6 in the bottom row and thus there are @xmath84 actual choices . in general , in the @xmath135-th row
counted from the bottom , there are @xmath136 columns , but @xmath137 are already occupied by @xmath6 s that are situated in rows below .
this leaves us with @xmath135 possibilities . in total , there are @xmath138 @xmath122s with @xmath29 rows and no @xmath5 . in order to give an indication as to why it is probably not easy to construct a bijection between @xmath4s and @xmath122s , we elaborate on the @xmath139 case in appendix [ k1 ] . as a further comment that is also related to the previous subsection ,
observe that in order to transform an odd @xmath1-triangle @xmath15 of order @xmath29 with @xmath51 into the corresponding @xmath122 of order @xmath50 , one simply has to replace all @xmath5 s along the left and right boundary by @xmath3 s .
if we delete the diagonals of an odd @xmath1-triangle of order @xmath29 with @xmath43 , then , by corollary [ char ] , we obtain triangular arrays of the following type . a _ dual alternating sign triangle _ ( @xmath140 ) of order
@xmath29 is a triangular array @xmath123 in which each entry is @xmath3 , @xmath6 or @xmath5 and the following conditions are fulfilled . 1 . [ dast-1 ] the non - zero entries alternate in each row and each column . 2 .
[ dast-2 ] all column sums are @xmath3 .
[ dast-3 ] the topmost non - zero entry of each column is @xmath6 ( if it exists ) .
the set of order @xmath29 @xmath140s is denoted by @xmath141 .
next we display all order @xmath125 @xmath140s .
@xmath142 it is possible to reconstruct the deleted diagonal entries , except for the rows that contain only zeros , referred to as @xmath3-rows in the following : * add a @xmath6 below the bottom entry . * if the leftmost non - zero entry of a row is @xmath5 ( resp . 1 ) , place a @xmath6 ( resp .
0 ) left of the leftmost entry of that row .
* if the rightmost non - zero entry of a row is @xmath5 ( resp . 1 ) ,
place a @xmath6 ( resp .
0 ) right of the rightmost entry of that row .
* in the case of @xmath3-rows , there are two choices of placing a @xmath6 on one end and a @xmath3 on the other
. therefore , @xmath143 the @xmath140s listed above correspond to @xmath144 @xmath1s of order @xmath145 with @xmath146 , respectively .
this is in accordance with theorem [ maxone ] as there are @xmath147 @xmath44s in a @xmath148 box .
we introduce another set of triangular arrays that is , on the one hand , equinumerous with the set of @xmath36 such that @xmath43 , and , on the other hand , contains all @xmath122s of order @xmath29 . a _ quasi alternating sign triangle _ ( @xmath149 ) of order @xmath29 is a triangular array @xmath123 in which each entry is @xmath3 , @xmath6 or @xmath5 and the following conditions are fulfilled . 1 . [ qast-1 ] the non - zero entries alternate in each row and column
[ qast-2 ] the row sums are @xmath6 for rows
@xmath150 , and @xmath3 or @xmath6 for row @xmath29 .
[ qast-3 ] the topmost non - zero entry in each column is @xmath6 ( if it exists ) .
the set of order @xmath29 @xmath149s is denoted by @xmath151 . to construct a bijection between @xmath151 and @xmath152 , recall from corollary [ char ] ( [ char-2 ] ) that the objects in the latter set correspond to triangular six - vertex configurations on @xmath153 such that , in each row , the leftmost and the rightmost vertical edge point upwards .
now we perform the following operations on such configurations : 1 . for the second vertex of each row ( excluding the first and the last row , which contain only degree @xmath6 vertices ) , we interchange the orientation of the bottom vertical edge with the orientation of the left horizontal edge incident with this vertex .
2 . for the penultimate vertex of each row (
excluding the first and the last row ) , we interchange the orientation of the bottom vertical edge with the orientation of the right horizontal edge incident with this vertex .
this leads to triangular six - vertex configurations of order @xmath29 such that , in each row , the leftmost horizontal edge points to the right and the rightmost horizontal edge points to the left with the exception of the rightmost horizontal edge in the penultimate row .
compare this to the triangular six - vertex configurations in corollary [ char ] ( [ char-1 ] ) which were shown to correspond to @xmath122s of order @xmath29 .
similarly , it can be shown that the present configurations correspond to @xmath149s of order @xmath29 .
if we perform only the first operation above , we obtain triangular six - vertex configurations that are equivalent to the following triangular arrays .
these arrays can be seen as a mixture of @xmath122s and @xmath140s .
a _ mixed alternating sign triangle _ ( @xmath154 ) of order @xmath29 is a triangular array @xmath123 in which each entry is @xmath3 , @xmath6 or @xmath5 and the following conditions are fulfilled . 1 . [ mast-1 ] the non - zero entries alternate in each row and column . 2
. [ mast-2 ] the column sums are @xmath3 for columns
[ mast-3 ] the first non - zero entry in each row and each column is @xmath6 ( if it exists ) .
similarly , it can be seen that @xmath124 is in bijection with the set of @xmath154s of order @xmath29 whose bottom entry is @xmath6 , and also in bijection with the set of @xmath154s of order @xmath29 whose @xmath29-th column has sum @xmath3 . here
are two related remarks . *
as @xmath156 , and @xmath124 is equinumerous with the set of @xmath8 @xmath4s , while @xmath151 is equinumerous with the set of @xmath44s in an @xmath46 box , there should exist a natural subset of the set of @xmath44s in an @xmath46 box that has the same cardinality as the set of order @xmath29 @xmath4s . *
the set of order @xmath29 @xmath149s can be partitioned according to the entry at the bottom and the central column sum : ( i ) the bottom entry is @xmath6 , ( ii ) the bottom entry is @xmath3 and the central column sum is @xmath3 , ( iii ) the bottom entry is @xmath3 and the central column sum is @xmath6 . each of the first two sets is actually equinumerous with @xmath124 .
the set of order @xmath29 @xmath149s that belong to the first or third class are equinumerous with odd @xmath1-triangles of order @xmath64 with the property that the entries of each row except for possibly the bottom row sum to @xmath6 , where possibly either or both of the boundary entries are disregarded .
indeed , such a @xmath149 is bijectively transformed into such an odd @xmath1-triangle by replacing each @xmath3 on the diagonals by @xmath5 if it is contained in a column with sum @xmath6 .
compare this to corollary [ char ] to see that the set of odd @xmath1-triangles @xmath15 of order @xmath64 with @xmath157 is a subset of the set of these @xmath1-triangles .
we have seen that the various classes of extreme @xmath1s under consideration in theorems [ maxmone][maxzero ] have an alternative interpretation in terms of certain orientations of triangular regions of the square grid .
the differences between the classes are due to the various conditions along the left , right and bottom boundary . in this section ,
we define a universal generating function in the language of physics a universal _
partition function_for all of these orientations of @xmath81 as well as two refinements according to the orientation of the bottom vertical edge .
these generating functions involve _ boundary parameters _ which are sufficiently general such that the generating function of each class of extreme @xmath1s is a specialization of the universal generating function or one of its refinements .
we assign a weight @xmath158 to each triangular six - vertex configuration @xmath159 .
this weight is the product of the vertex weights to be defined next .
the weight of an individual vertex depends besides a global variable @xmath160on the _ orientations of the surrounding edges _ , on a _ label _ which is assigned to the vertex , and , for bulk vertices , on the _ position of the label_. we define @xmath161 and we use this notation throughout the rest of this paper .
the vertex weights are provided in table [ weights ] , with the exception of the degree @xmath6 vertices , which always have weight @xmath6 .
the label of the vertex is denoted by @xmath162 , and we assume in this table that for bulk vertices , the label is situated in the south - west corner , as follows .
@xmath163 also elsewhere , if the position of the label of a degree @xmath79 vertex is not indicated , it is assumed to be in the south - west corner . if the label appears in some other corner , one has to rotate the diagram until the label is in the south - west corner , and then determine the weight .
the left boundary weights depend on the _ left boundary constants _
@xmath164 , and the right boundary weights depend on the _ right boundary constants _ @xmath165 .
@xmath166{0ex}{3.5ex } \text{bulk weights } & \text{left boundary weights } & \text{right boundary weights } & \text{matrix entry } \\
\hline \hline\rule[-0.6ex]{-0.4ex}{3.5ex } { { \operatorname{w}}({\raisebox{-1.5mm}{\scalebox{0.2}{\includegraphics{1}}}},u)}=1 & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{1l } } } } , u)}= \frac{\beta_l \ , q u + \gamma_l \ , \bar q \bar u}{\sigma(q^2 ) } & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{1r } } } } , u ) } = \frac{\beta_r \ , q \bar u + \gamma_r \ , \bar q u}{\sigma(q^2 ) } & 1 \\[1.5 mm ] { { \operatorname{w}}({\raisebox{-1.5mm}{\scalebox{0.2}{\includegraphics{m1 } } } } , u)}=1 & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{lm1 } } } } , u ) } = \frac{\gamma_l \ , q u + \beta_l \ , \bar q \bar u}{\sigma(q^2 ) } & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{rm1 } } } } , u ) } = \frac{\gamma_r \ , q \bar u + \beta_r \ , \bar q u}{\sigma(q^2 ) } & -1 \\[1.5 mm ] { { \operatorname{w}}({\raisebox{-1.5mm}{\scalebox{0.2}{\includegraphics{sw } } } } , u ) } = \frac{\sigma(q^2 u)}{\sigma(q^4 ) } & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{lin } } } } , u ) } = \alpha_l \ , \frac{\sigma(q^2u^2)}{\sigma(q^2 ) } & & 0 \\[1.5 mm ] { { \operatorname{w}}({\raisebox{-1.5mm}{\scalebox{0.2}{\includegraphics{ne } } } } , u)}= \frac{\sigma(q^2 u)}{\sigma(q^4 ) } & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{lout } } } } , u ) } = \delta_l \ , \frac{\sigma(q^2u^2 ) } { \sigma(q^2 ) } & & 0 \\[1.5 mm ] { { \operatorname{w}}({\raisebox{-1.5mm}{\scalebox{0.2}{\includegraphics{se } } } } , u ) } = \frac{\sigma(q^2 \bar u)}{\sigma(q^4 ) } & & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{rin } } } } , u ) } = \alpha_r \ , \frac{\sigma(q^2 \bar u^2)}{\sigma(q^2 ) } & 0 \\[1.5 mm ] { { \operatorname{w}}({\raisebox{-1.5mm}{\scalebox{0.2}{\includegraphics{nw } } } } , u ) } = \frac{\sigma(q^2 \bar u)}{\sigma(q^4 ) } & & { { \operatorname{w}}({\raisebox{0mm}{\scalebox{0.2}{\includegraphics{rout } } } } , u ) } = \delta_r \ , \frac{\sigma(q^2 \bar u^2)}{\sigma(q^2 ) } & 0 \\[1.5 mm ] \hline \end{array}\ ] ] the following observations are crucial . * for bulk vertices ,
the weight is unchanged when reversing the orientation of all edges . * for bulk vertices , rotation of the configuration or the label by @xmath167
is equivalent to replacing the label @xmath162 by @xmath168 . *
reflection of a local configuration in the vertical symmetry axis is equivalent to replacing @xmath162 by @xmath168 and , in addition for boundary weights , interchanging @xmath169 and @xmath170 in the boundary constants .
the _ label _ of a vertex in @xmath81 is determined as follows : the path in @xmath81 that is induced by the sequence of vertices corresponding to the entries in is associated with the _ spectral parameter _
@xmath171 , see also figure [ odasasmfig ]
. the label of a vertex is then the product of the spectral parameters associated with the paths which contain that vertex . in the case of bulk vertices , there are two such paths , while the path is unique for boundary vertices . _ example . _
the left configuration in figure [ implies ] has the weight @xmath172 in this subsection , we demonstrate that the weights introduced in the previous subsection satisfy certain local equations .
the most prominent local equation is the _ yang baxter equation _ , which is stated first .
the schematic diagrams in this subsection have to be interpreted as follows , see for instance .
each diagram is a graph which involves a number of external edges , for which only one endpoint is indicated with a bullet .
close to the other endpoint of such an edge , we place an @xmath173 which indicates its orientation , i.e.,``@xmath174 '' or `` @xmath175 '' , relative to the indicated endpoint of the external edge . a diagram represents the generating function of all orientations of the graph such that the external edges have the prescribed orientation , degree @xmath79 vertices are balanced , and the vertex weights are as given in table [ weights ] , where the parameter near a vertex indicates its label .
[ yang ] if @xmath176 and @xmath177 , then @xmath178 the theorem can be proven by verifying that , for each choice of orientations of the external edges , is either trivial or reduces to a simple identity satisfied by the bulk weights in table [ weights ] . to deal with triangular six - vertex configurations
, we also need left and right forms of the reflection equation .
the reflection equation was introduced ( and applied to six - vertex model bulk weights ) by cherednik ( * ? ? ?
* eq . ( 10 ) ) , with important further results being obtained by sklyanin @xcite . [ re ]
suppose @xmath179 and the bulk weights are as given in table [ weights ] .
then @xmath180 for all @xmath181 if and only if there exist parameters @xmath182 independent of @xmath162 and a function @xmath183 such that @xmath184 and @xmath185 similarly , if @xmath186 and the bulk weights are chosen as in table [ weights ] , then @xmath187 for all @xmath181 if and only if @xmath188 are chosen up to multiplication by a function @xmath189as in table [ weights ] .
the result was obtained , independently , by de vega and gonzlez - ruiz ( * ? ? ?
* eq . ( 15 ) ) , and ghoshal and zamolodchikov ( * ? ? ?
* eq . ( 5.12 ) ) .
we also provide a proof below .
first we show that every solution of is of the form as given in the statement of the theorem .
set @xmath190 , @xmath191 in and obtain ( after subtraction of an identical term from each side ) @xmath192 the equation shows in particular that @xmath193 if and only if @xmath194 .
we first assume @xmath195 and @xmath196 .
we divide by @xmath197 and deduce @xmath198 after plugging in the definitions for the bulk weights and setting @xmath199 , @xmath200 , we can deduce that @xmath201 with @xmath202 and @xmath203 .
since these are both parameters independent of @xmath162 and at least one of them is non - zero , we can deduce that @xmath204 and @xmath205 are of the form stated in the theorem . in
, we now choose @xmath206 , @xmath207 and obtain @xmath208 in this identity , every quantity is known ( in the case of @xmath204 and @xmath205 up to @xmath209 ) , except for @xmath210 and @xmath211 .
the identity thereby becomes @xmath212 from which we see that @xmath210 is also of the form given in the theorem . in order to obtain the expression for @xmath213 ,
choose @xmath214 and @xmath215 in , which gives @xmath216 if , on the other hand , @xmath193 and @xmath194 , we choose @xmath217 and the assertion also follows simply from .
conversely , it is straightforward to check that the solution stated in the theorem satisfies for all @xmath181 .
the right reflection equation follows from the left reflection equation by substituting @xmath218 .
there are two additional simple rules that are necessary . for @xmath219
, we have @xmath220 and @xmath221 where @xmath222 , @xmath223 and @xmath224 is the kronecker delta .
the universal partition function is the generating function of all permissible orientations of @xmath81 with respect to the weight function we have defined , that is @xmath225 where the sum is over all orientations @xmath159 of @xmath81 in which each bulk vertex has two incoming and two outgoing edges and the top edges point up .
moreover , @xmath226 and @xmath227 denote the restrictions of this sum to configurations in which the bottom vertical edge points up or down , respectively . in ( * ? ? ?
* eq . ( 24 ) ) , it was shown that the two refinements are expressible in terms of the full partition function , as follows .
[ updown ] the universal partition function and its refinements satisfy @xmath228 in that paper we assumed more special boundary weights , but the proof of the lemma is independent of the form of the boundary weights , since the argument involves only vertex weights that depend on @xmath229 , and this parameter appears in none of the boundary weights .
next , we introduce three specializations of the spectral parameters @xmath230 , the global parameter @xmath160 and the boundary parameters @xmath231 , for which @xmath232 counts extreme @xmath1s in theorems [ maxmone ] , [ maxone ] and [ maxzero ] , respectively . in sections
[ ast][qoosasm ] , we then derive determinant or pfaffian expressions for these specializations of the partition function.s using the weights in table [ weights ] .
in contrast to @xmath233 , we are able to derive an expression for the universal partition function in this case . ]
first observe that if we specialize @xmath234 for all @xmath235 and @xmath236 , then all bulk weights are equal to @xmath6 . by corollary [ char ]
, we need to have @xmath237 if @xmath238 . however , it turns out to be useful to introduce an additional global parameter @xmath239 that is set to @xmath6 to prove theorem [ maxmone ] .
thus , it is sufficient to have @xmath240 as well as @xmath241 this is satisfied if we assign the following values to the boundary parameters .
note that this is true even if we do not specialize @xmath160 .
this is also the case for the two other specializations . @xmath242
the explicit boundary weights are therefore as follows .
@xmath243 we denote this specialization of the universal partition function as @xmath244 and record that @xmath245 by corollary [ char ] , we require @xmath246 and the bottom vertical edge needs to point up , that is we need to work with @xmath247 . again , we introduce a parameter @xmath239 that is set to @xmath6 for proving theorem [ maxone ] , and require @xmath248 as well as @xmath249 we choose the boundary parameters as follows .
@xmath250 in this case , the explicit boundary weights are @xmath251 we denote this specialization of the universal partition function as @xmath252 and obtain that @xmath253 by corollary [ char ] , we need to have @xmath254 and so it is natural to choose the following .
@xmath255 here , the left boundary weights are @xmath256 in this case , the left boundary weight in row @xmath135 of a configuration with non - zero weight is @xmath257 , while the right boundary weight in row @xmath135 is @xmath258 .
this implies that the universal partition function has the factor @xmath259 in this specialization .
we cancel this factor and denote the reduced specialization as @xmath260 .
therefore , @xmath261 note that this normalization of the universal partition function could alternatively be achieved by taking the functions @xmath183 and @xmath189 in theorem [ re ] to be @xmath262 and @xmath263 .
kuperberg used in @xcite analogous specializations to study @xmath264s and @xmath56s of order @xmath11 .
if we require @xmath265 to count all odd - order @xmath1s as was done in @xcite , then we may choose @xmath266 and @xmath267 . here
, the left boundary weight in row @xmath135 is divisible by @xmath268 , while the right boundary weight is divisible by @xmath269 .
thus we may divide this specialization by @xmath270 and still obtain a laurent polynomial in @xmath271 .
( alternatively , this normalization could be achieved by using certain choices for the functions @xmath183 and @xmath189 in theorem [ re ] . )
this normalization was used in @xcite .
in ( * ? ? ? * proposition 12 ) , it was shown how theorems [ yang ] and [ re ] imply the symmetry of the @xmath1 specialization ( see subsection [ dasasm - spec ] ) of the order @xmath29 partition function in @xmath272 . since these theorems also hold for the more general weights given in table [ weights ] , the symmetry generalizes to the universal partition function . by lemma [ updown ]
, we have symmetry also for the two refined partition functions . [ sym ] the partition function @xmath233 , and its two refinements
@xmath273 are symmetric in @xmath274
. moreover , we have the following invariance .
[ inv ] the partition function @xmath233 and its two refinements @xmath273 are invariant when simultaneously replacing @xmath275 with @xmath276 and interchanging left and right boundary constants , i.e. , @xmath277 .
this follows from reflecting the configuration in the vertical symmetry axis as the reflection causes for each individual vertex the replacement of the label by its reciprocal and , for boundary weights , the interchanging of respective left and right boundary constants .
finally , @xmath278 are laurent polynomials in @xmath271 that are even in each @xmath279 , @xmath280 .
we need the following notion : suppose @xmath281 is a laurent polynomial in @xmath162 , and @xmath282
. then @xmath283 is the _ degree _ of the laurent polynomial and @xmath284 is its _
order_. [ even ] 1 .
[ even-1 ] for each @xmath285 , the partition function @xmath233 and its two refinements @xmath273 are even in @xmath279 .
[ even-2 ] the partition function @xmath233 and its two refinements @xmath273 are laurent polynomials in @xmath230 of degree at most @xmath29 and order at least @xmath286 in @xmath229 , and of degree at most @xmath287 and order at least @xmath288 in @xmath289 , for each @xmath285 . _ : _ by theorem [ sym ] , it suffices to show the assertion for @xmath290 .
we show that the contribution of each configuration to the partition function is even in @xmath291 .
distinguish between the cases whether or not the unique @xmath6 in the top row of the @xmath1 is on the ( left or right ) boundary .
clearly , only the weights of the top bulk vertices and the two top boundary vertices can involve the spectral parameter @xmath291 , and if the @xmath6 is on the left boundary , then the contribution of the first row to the weight is@xmath292 while the contribution is @xmath293 if the @xmath6 is on the right boundary .
both expressions are obviously even in @xmath291 .
if the unique @xmath6 in the top row is not on the boundary , then the contribution of the top bulk vertices and the two top boundary vertices is the product of @xmath294 and an even number ( @xmath295 ) of @xmath296 s and @xmath297 s , @xmath298 , and this product is also even in @xmath291 . _ : _ the bounds for the degree and order in @xmath279 , @xmath285 , follow from the discussion in the proof of . as for the degree and order in @xmath229 , note that there are precisely @xmath29 vertices of @xmath81 whose
weight can involve @xmath229 ( i.e. , the bulk vertices of the central column ) .
the degree of the weight in @xmath229 of such a vertex is at most @xmath6 and the order is at least @xmath5 . in the following lemma
we show that the evaluation of the universal partition function @xmath299 at @xmath300 is expressible in terms of the order @xmath64 universal partition function .
[ eval ] for any @xmath301 , @xmath302 where @xmath303 concerning the two refinements , we have @xmath304 we prove and illustrate the proof with the help of the case @xmath305 . by definition ,
the left - hand side is the generating function of the following graph where @xmath300 .
the weight of the top bulk vertex in the central column is @xmath306 . using and observing that the local configurations in the left half of the top row are now forced ,
we see that this is equal to the generating function of the following graph .
the @xmath29 vertices with fixed local configuration in the left half of the top row contribute @xmath307 to the weight .
we delete these vertices and move the right half of the graph down .
now we apply the yang
baxter equation to the vertices with label @xmath308 , @xmath309 and to the right vertex with label @xmath310 ( indicated with yellow in our example ) .
this is permissible since @xmath300 .
we repeatedly apply the yang
baxter equation until we reach the right boundary , and then apply the right reflection equation .
we then untangle the crossing and gain the factor @xmath311 . in our example , the procedure is as follows .
next we apply the yang
baxter equation to the vertices with label @xmath312 , @xmath309 and to the right vertex with label @xmath313 and , with repeated applications of , move to the right as much as possible
unless this triangle is already at the right boundary ( as in our example ) .
we then use the right reflection equation and afterwards the yang
baxter equation until we reach the top .
we untangle and gain the factor @xmath314 . in our example
, we perform the following .
we continue in this manner and obtain in total the factor @xmath315 .
we generating function of the resulting graph can be expressed as @xmath316 the proof of is analogous , and then follows from and using lemma [ updown ] . in (
* proposition 15 ) , we have shown the following identity for the @xmath1 specialization of the order @xmath29 partition function . @xmath317 the proof is based on , and , and can thus be generalized to our more general weights . the identity is equivalent to lemma [ eval ] , due to lemma [ updown ] . [ un ]
let @xmath29 be a positive integer .
a laurent polynomial @xmath318 in @xmath229 over the field @xmath319 is equal to @xmath233 if and only if the following conditions are fulfilled . 1 . [ 1degree ]
the degree of @xmath318 in @xmath229 is no greater than @xmath29 and the order is no smaller than @xmath286 .
[ 1sym ] @xmath318 is symmetric in @xmath272 .
[ 1inv ] @xmath318 is invariant when simultaneously replacing @xmath275 with @xmath320 and interchanging left and right boundary constants .
4 . [ 1even ] @xmath318 is even in @xmath279 , for all @xmath285 .
[ 1eval ] @xmath318 satisfies the identity obtained from by replacing on the left - hand side @xmath321 with @xmath322 .
analogous characterizations for @xmath323 and @xmath324 are obtained by replacing in with and , respectively .
we consider @xmath232 and @xmath318 as laurent polynomials in @xmath229 , and , by proposition [ even ] and , it suffices to show that they coincide at @xmath13 evaluations .
now one evaluation ( at @xmath325 ) is given by .
using and , as well as the facts that @xmath326 are odd in @xmath291 , while @xmath327 are even , we obtain the evaluation at @xmath328 as follows .
@xmath329 using , we also obtain the evaluations at @xmath330 as follows .
@xmath331 these two identities are of course also valid if we replace @xmath322 by @xmath321 as the partition function has all the properties listed in the statement of the theorem by theorem [ sym ] , proposition [ inv ] , proposition [ even ] and lemma [ eval ] . by the symmetry in @xmath332 ,
@xmath291 can be replaced by any @xmath279 , @xmath333 , and so we have in total @xmath11 evaluations in @xmath229 , namely at @xmath334 , @xmath285 .
the proofs for @xmath335 and @xmath336 are similar .
this section provides preparations for alternative proofs of some results that avoid lemma [ eval ] , and it can therefore also be omitted when reading the article .
[ lem_eval1 ] let @xmath29 be a positive integer .
for @xmath337 , @xmath338 the corresponding identities also hold for the refined partition functions @xmath339 .
the possible local configurations for the top right boundary vertex are @xmath340 and @xmath115 . however , as @xmath341 , this local configuration must be @xmath340 , which causes a fixing of the top row . more precisely , the local configurations of the top bulk vertices are @xmath342 , and @xmath113 for the top left boundary vertex .
[ lem_eval2 ] let @xmath343 . for @xmath344 , @xmath345
the corresponding identities also hold for the refined partition functions @xmath346 .
this identity appeared before for the @xmath1 specialization of the universal partition function in ( * ? ? ? * proposition 14 ) .
the proof is based on and , and since these rules are also valid for our more general weights , is true in general .
[ u1 ] let @xmath343 and @xmath347 be indeterminates .
suppose we consider a specialization of @xmath232 where @xmath348 and @xmath349 such that * @xmath232 has a zero at @xmath350 , and * left boundary constants are transformed into corresponding right boundary constants when replacing @xmath239 by @xmath351 .
a laurent polynomial @xmath352 in the variables @xmath332 with coefficients in @xmath353 is equal to this specialization of @xmath233 if and only if the following conditions are fulfilled . 1 . [ 2degree ]
the degree of @xmath352 in @xmath354 is no greater than @xmath287 and the order is no smaller than @xmath288 .
[ 2sym ] @xmath352 is symmetric in @xmath272 .
[ 2inv ] @xmath352 is invariant under the transformation @xmath355 .
[ 2even ] @xmath352 is even in @xmath279 , for all @xmath285 .
[ 2eval ] @xmath352 satifies the identities obtained from and by replacing on the left - hand sides @xmath232 by @xmath352 and considering the same specializations of the boundary constants and @xmath229 on the right - hand sides . 6 .
[ 2extra ] @xmath352 has a zero at @xmath356 .
analogous characterizations for @xmath323 and @xmath324 are obtained by replacing and with the respective identities .
the partition function @xmath232 satisfies properties by proposition [ even ] , theorem [ sym ] , proposition [ inv ] , and lemmas [ lem_eval1 ] and [ lem_eval2 ] , and property by assumption .
an even laurent polynomial of degree at most @xmath287 and order at least @xmath288 is uniquely determined by @xmath55 evaluations where we need to guarantee that there is no pair of evaluations which differ only in sign .
we have @xmath357 evaluations ( at @xmath358 , @xmath359 ) if we combine the evaluation obtained from with and , and @xmath79 evaluations ( at @xmath360 ) if we combine the two evaluations obtained from with .
the additional evaluation is provided by .
the main purpose of this section is to provide the proof of theorem [ maxmone ] and generalizations .
[ astpartfunct ] we have the following determinant formula for the partition function @xmath361 .
@xmath362 1 - \frac{\sigma(\bar u_{n+1 } u_j)}{\sigma(\bar p u_j)},&i = n+1\end{cases}\right)\end{gathered}\ ] ] we use theorem [ un ] .
the proof is by induction with respect to @xmath29 .
the case @xmath363 is easy to verify . to show that is indeed a laurent polynomial in @xmath229 , observe that @xmath364 1 - \frac{\sigma(\bar u_{n+1 } u_j)}{\sigma(\bar p u_j)},&i = n+1\end{cases}\right)\ ] ] is a laurent polynomial in @xmath229 .
it is divisible by @xmath365 , as the determinant vanishes if we set @xmath366 , for @xmath367 , since then the @xmath235-th column coincides with the @xmath368-st column of the matrix . : observe that @xmath369 1 - \frac{\sigma(\bar u_{n+1 } v_j)}{\sigma(\bar p v_j)},&i = n+1\end{cases}\right)\end{gathered}\ ] ] is a laurent polynomial in the variables @xmath370 that is antisymmetric in @xmath272 and in @xmath371 .
it is also even in each @xmath279 , @xmath280 , and odd in each @xmath372 , @xmath373 , and thus divisible by @xmath374 . after setting @xmath375 ,
it is also divisible by @xmath376 , since then the bottom right corner entry of the matrix is @xmath6 . ]
we check that has the desired properties from theorem [ un ] . _ : _ to deduce the bounds on the degree and order of as a laurent polynomial in @xmath229 , use the leibniz formula for the determinant and observe that , after multiplying each summand with the prefactor and cancelling as much as possible , we obtain a sum of rational functions .
each numerator has degree @xmath377 or @xmath130 ( depending on whether or not the entry in the bottom right corner of the matrix is involved ) and order @xmath378 or @xmath379 .
the denominators have degree @xmath29 and order @xmath286 , and since we already know that is a laurent polynomial in @xmath229 , the bounds follow .
_ , and : _ the expression is obviously symmetric in @xmath332 , and also even in @xmath279 , for each @xmath380 .
moreover , it is invariant under the transformation @xmath381 .
note that interchanging respective left and right boundary constants in the @xmath122-specialization ( see subsection [ ast - spec ] ) is equivalent to the replacement @xmath382 .
_ : _ let @xmath383 denote .
then we need to show the following .
@xmath384 for this purpose , multiply the @xmath368-st column of the matrix in @xmath385 with the prefactor @xmath386 . after taking @xmath300 ,
the @xmath368-st column is @xmath387 .
now expand the determinant with respect to this column to reduce the computation to the determinant of an @xmath8 matrix .
after moving the first column to the right , note that the first @xmath64 rows of this matrix match those of the matrices in @xmath388 and @xmath389 .
take the linear combination of the last rows in the right - hand side according to the prefactors and compare it with the last row and the prefactor in the left - hand side .
taking @xmath390 in , the last row of the determinant becomes @xmath391 , and we obtain the following simplified formula .
[ astpartfunct1 ] we have the following formula for @xmath392 .
@xmath393 next we provide a sketch of an alternative proof of corollary [ astpartfunct1 ] which avoids theorem [ astpartfunct ] and lemma [ eval ] , the latter being probably the most complicated ingredient in the proof of theorem [ astpartfunct ] .
this proof is also more in line with kuperberg s approach in @xcite as it uses a characterization of the partition function in terms of evaluations in @xmath354 .
first we need to verify that @xmath394 by the symmetry of theorem [ sym ] , it suffices to show that @xmath395 .
consider the bottom bulk vertex @xmath396 in the central column and assume @xmath397 .
its label is @xmath398 , and since @xmath399 , there are only four possible local configurations around @xmath396 that contribute a non - zero weight ( which is in fact @xmath6 in all cases as @xmath400 . moreover , as @xmath401 , there are only two possibilities ( namely @xmath402 ) for the right boundary vertex @xmath403 adjacent to @xmath396 .
it turns out that the weights of the remaining configurations cancel in pairs : fix a configuration and obtain another configuration by reversing the orientation of the bottom vertical edge incident with @xmath396 and the right horizontal edge incident with @xmath396 .
then only the weight of @xmath403 has changed : it is @xmath404 , if the horizontal edge points to the right , and it is @xmath405 if the horizontal edge points to the left .
now we may use theorem [ u1 ] and induction with respect to @xmath29 .
the cases @xmath406 are easy to verify .
( we interpret empty products and empty determinants as @xmath6 . ) to show that is a laurent polynomial in @xmath332 , proceed as in the footnote of the proof of theorem [ astpartfunct ] .
_ : _ we use the leibniz rule for the determinant in . the degree and order of the numerator in @xmath291 of each summand
after multiplication with the prefactor and cancelling are @xmath407 and @xmath408 , respectively , while the degree and order of the denominator are @xmath357 and @xmath409 , respectively . _ , , ,
and : _ the expression is symmetric in @xmath272 , invariant under the transformation @xmath410 , and even in @xmath279 , @xmath285 .
the identities obtained from lemmas [ eval1 ] and [ eval2 ] can be verified straightforwardly .
it is also evident that the expression in has a zero at @xmath350 .
a comparison with the partition function of all @xmath4s shows a close relation to the @xmath122 partition function at @xmath348 : let @xmath411 denote the partition function of @xmath8 @xmath4s ( @xmath272 and @xmath412 are the horizontal and vertical spectral parameters , respectively , and we use the bulk weights given in table [ weights ] ) . then @xmath413 to see this ,
consult for instance ( * ? ? ?
* theorem 10 ) : set @xmath414 in kuperberg s formula and divide by @xmath415 to take the difference in the choice of vertex weights into account .
a determinant formula for the partition function of the six - vertex model on an @xmath8 grid with domain wall boundary conditions that is equivalent to the one above was first derived by izergin ( * ? ? ?
* eq . ( 5 ) ) , using results of korepin @xcite .
corollary [ astpartfunct1 ] now implies that @xmath416 in particular , now implies that there is the same number of order @xmath29 @xmath4s as there is of order @xmath29 @xmath122s , since @xmath417 is the number of order @xmath29 @xmath4s .
next we specialize @xmath238 and obtain an expression involving _ schur functions _
we use the well - known determinant formula @xmath419 where @xmath420 , and if the partition has less than @xmath29 parts , we add the appropriate number of zero parts to the partition . [ astschurtheo ]
we have the following schur function expression for @xmath392 at @xmath238 .
@xmath421 use and ( * ? ? ?
* theorem 2.4 ( 1 ) , 2nd eq . ) .
we note that the schur function expression for the asm partition function at @xmath238 was obtained simultaneously by okada and by stroganov ( * ? ? ?
* eq . ( 17 ) ) .
let us point out that character formulas for the specializations of @xmath392 at @xmath422 are also provided in ( * ? ? ?
* theorem 2.4 ( 1 ) ) .
these specializations correspond to the @xmath423-enumerations of @xmath4s ( or @xmath122s , see also subsection [ xenum ] ) for @xmath424 , respectively . in a forthcoming paper @xcite
, we will see that factorizes .
more precisely , it will be shown that @xmath425 where the notation @xmath426 and @xmath427 for the orthogonal characters is the same as that is used in @xcite .
these factorizations are very similar to those obtained by ciucu and krattenthaler in that paper .
the inversion number of an order @xmath29 @xmath4 @xmath428 is defined as @xmath429 if @xmath15 is a permutation matrix , @xmath430 is the number of inversions of the permutation @xmath431 , where @xmath432 is the column for the unique @xmath6 in row @xmath135 .
the inversion number of @xmath4s was first defined in ( * ? ? ?
* eq . ( 18 ) ) , where it was referred to as the number of positive inversions .
another generalization of the inversion number of permutations is obtained if , instead of summing over all @xmath433 , we sum over all @xmath434 , and this statistic , which in fact evaluates to @xmath435 , appeared in @xcite . in terms of the six - vertex model , we have @xmath436 where @xmath437 denotes the six - vertex configuration of @xmath15 .
this is because @xmath438 i.e. , @xmath430 is the number of entries @xmath439 of @xmath15 such that the sum of entries in the same column above @xmath439 ( not including @xmath439 ) is @xmath6 and the sum of entries in the same row to the left of @xmath439 ( including @xmath439 ) is @xmath6 .
it is not hard to see that also @xmath440 or , equivalently , @xmath441 .
this implies @xmath442 we generalize this version of the inversion number to @xmath122s straightforwardly : suppose @xmath443 .
then let @xmath444 where @xmath445 is the triangular six - vertex configuration associated with @xmath446 .
if @xmath428 is the @xmath21 @xmath1 corresponding to @xmath446 , this can also be written as follows .
@xmath447 this expression can be simplified as we show next . in a @xmath1 @xmath428 of order @xmath13 with @xmath26 ,
we have @xmath448 and @xmath449 if @xmath380 , since @xmath450 by corollary [ char ] ( [ char-1 ] ) , and so can also be written as @xmath451 let @xmath452 be the corresponding @xmath122 , then this is @xmath453 where in the latter expression both sums range over all @xmath454 such that @xmath455 is defined . after observing that @xmath456 which is true since @xmath457 and there are @xmath29 columns whose entries sum to @xmath6 ,
while all other column sums are @xmath3 , this implies @xmath458 in particular , we can conclude that the inversion number is an integer .
[ maxmone - gen ] the joint distribution of the statistics @xmath34 and @xmath41 on the set @xmath32 is equal to the joint distribution of the statistics @xmath35 and @xmath39 on the set @xmath124 , i.e. , for all integers @xmath132 and @xmath135 we have @xmath459 define the following complementary inversion numbers for @xmath460 and @xmath443 : @xmath461 as there are @xmath29 more @xmath6 s in an @xmath4 ( resp .
@xmath122 ) of order @xmath29 than @xmath5 s , we have @xmath462 and so it suffices to show @xmath463 the left hand side can be studied using the partition function @xmath464 .
indeed , we have @xmath465 since the ( bulk ) weights for the @xmath4 partition function @xmath411 are chosen as in table [ weights ] and the label of the vertex in row @xmath135 and column @xmath235 is @xmath466 .
now it is important to note that @xmath467 and @xmath468 are algebraically independent if @xmath347 are ( algebraically independent ) indeterminates .
indeed , let @xmath469 be a polynomial in @xmath470 and @xmath283 its total degree .
then @xmath471 is a laurent polynomial in @xmath239 of degree @xmath472 with leading coefficient @xmath473 and thus non - zero . on the @xmath122 side , we consider the specialization @xmath474 . by
, the sum in the generating function @xmath474 can be taken over all odd @xmath1-triangles @xmath475 of order @xmath29 with @xmath476 , and by , all left boundary weights are @xmath477 , while all right boundary weights are @xmath478 . hence , @xmath479 and the assertion now follows from and .
a statistic that was extensively studied for @xmath4s is the position of the @xmath6 in the top row , see for instance @xcite . in this subsection
, we study the analogous statistic on @xmath122s and derive a formula for the number of @xmath122s of order @xmath29 that have the @xmath6 in the top row in a prescribed position in terms of the doubly refined enumeration of @xmath4s with respect to the positions of the @xmath6 s in the top row and leftmost column .
explicit formulas for this doubly refined enumeration of @xmath4s can in turn be derived from results in @xcite , see also @xcite . for an @xmath122 @xmath446 ,
let @xmath480 and , for an @xmath4 @xmath15 , let @xmath481 we define two generating functions as follows .
@xmath482 for example , @xmath483 and @xmath484 . by setting @xmath485 in
, it can be shown ( where we leave the details of the derivation to the reader ) that @xmath486 where @xmath487 denotes the number of @xmath8 asms .
this implies @xmath488 with @xmath489 and @xmath490 .
the recursion can be solved and we obtain @xmath491 if @xmath492 denotes the number of @xmath8 @xmath4s that have the @xmath6 in the top row in column @xmath235 and the @xmath6 in the leftmost column in row @xmath135 , then @xmath493 each left or right boundary entry of an @xmath122 @xmath446 is either @xmath3 or @xmath6 .
let @xmath494 denote the number of @xmath6 s on the left boundary , and @xmath495 the number of @xmath3 s on the right boundary that are contained in a column with sum @xmath6 .
define the statistic @xmath496 for @xmath443 .
we have numerical evidence that , for any integers @xmath29 and @xmath497 , @xmath498 in a forthcoming paper , we will study this statistic and derive related constant term identities .
interestingly , data suggests that the joint distribution of @xmath499 and @xmath500 on @xmath124 is not the same as the joint distribution of @xmath501 and the position of the unique @xmath6 in the top row of an @xmath4 on @xmath32 . the following weighted enumeration of @xmath4s is referred to as the @xmath423-enumeration : @xmath502 if @xmath503 , then we obviously obtain @xmath134 since @xmath4s with no @xmath5 are precisely permutation matrices .
besides @xmath504 , the numbers are also round for @xmath505 ( see @xcite ) and @xmath506 ( see @xcite ) .
theorem [ maxmone ] states that the @xmath423-enumeration of @xmath122s is the same as the @xmath423-enumeration of @xmath4s for any @xmath423 , hence we obtain the following .
[ 23 ] the @xmath84-enumeration of order @xmath29 @xmath122s is @xmath507 , while the @xmath125-enumeration of order @xmath13 @xmath122s is @xmath508 and @xmath509 for order @xmath287 @xmath122s .
curiously , as demonstrated in @xcite ( see also ( * ? ? ?
* remark 4.3 ) ) , the @xmath84-enumeration of @xmath4s can be translated into the plain enumeration of _ perfect matchings _ of certain regions of the square grid ( or , equivalently , into the plain enumeration of the domino tilings of the _ aztec diamond _ ) .
similarly , we may now translate our result on the @xmath84-enumeration of @xmath122s into a matching enumeration result . in order to do so ,
define for each integer @xmath510 a graph @xmath511 as indicated in figure [ m5 ] ( left ) .
it consists of @xmath29 rows of centered sequences of tilted squares ( shaded in our drawing and referred to as diamonds in the following ) of lengths @xmath512 , from top to bottom .
the bottom vertices of the columns ( except for the middle column ) play a special role and these vertices are denoted by their column position , counted from the left . note that @xmath513 is bipartite and the vertices @xmath514 all lie in the same vertex class .
there are @xmath64 more vertices in this class than in the other class , and hence , in order to obtain a graph that could possibly possess a perfect matching , we may delete @xmath64 vertices in @xmath515 from @xmath513 .
there is an example of such a perfect matching in figure [ m5 ] ( right ) , where we delete vertices @xmath516 .
now , on the @xmath122-side , since each of the @xmath29 rows of an order @xmath29 @xmath122 has sum @xmath6 , and each of the @xmath130 columns has sum @xmath3 or @xmath6 , there must be precisely @xmath64 columns that have sum @xmath3 .
let @xmath517 and denote by @xmath518 the subset of @xmath124 , where the columns with sum @xmath3 are the columns @xmath519 .
( note that this set is obviously empty if @xmath520 for a @xmath235 . ) in the example in figure [ m5 ] ( right ) , we indicate a surjection from the set of perfect matchings of @xmath521 onto @xmath518 ( see also ( * ? ? ?
* remark 4.3 ) ) : each diamond corresponds to an entry of the @xmath122 , and this entry is in fact the number of matching edges on the boundary of the diamond . to compute the size of the preimage of a given @xmath122 under this map ,
observe that for each @xmath6 of the @xmath122 , there are obviously two possibilities to arrange the two matching edges on the boundary of the respective diamond . in summary , @xmath522 as @xmath523 , corollary [ 23 ] now implies the following .
[ match ] for any @xmath301 , @xmath524 the number of perfect matchings of a tilted square region of the square grid of order @xmath29 ( dual to the aztec diamond of order @xmath29 ) is also @xmath525 , since the plain enumeration of the perfect matchings of this region is up to the factor @xmath526 equal to the @xmath84-enumeration of order @xmath29 @xmath4s as mentioned above .
a bijection between the set of perfect matchings of this region and the set of perfect matchings of the family of graphs @xmath527 might give a hint on the bijection between @xmath4s and @xmath122s .
the main purpose of this section is to provide a proof of theorem [ maxone ] and generalizations .
[ dastpartfunct ] we have the following determinant formula for the partition function @xmath528 .
@xmath529 we use theorem [ un ] .
the proof is by induction with respect to @xmath29 .
the case @xmath363 is easy to verify . to show that is indeed a laurent polynomial in @xmath229 ,
observe that @xmath530 is a laurent polynomial in @xmath229 .
it is divisible by @xmath365 , as the determinant vanishes if we set @xmath366 , for @xmath367 , since then the @xmath235-th column coincides with the @xmath368-st column of the matrix up to sign .
we check the properties from theorem [ un ] . _ : _ to deduce the bounds on the degree and order as a laurent polynomial in @xmath229 , use the leibniz formula for the determinant and observe that , after multiplying each summand with the prefactor and expanding , we obtain a sum of rational functions where each numerator has degree @xmath377 and order @xmath378 , and each denominator has degree @xmath29 and order @xmath286 . _ , and : _ the expression is obviously symmetric in @xmath332 , invariant under the transformation @xmath531 and even in @xmath279 , @xmath285 . note that interchanging respective left and right boundary constants in the @xmath149-specialization ( see subsection [ dast - spec ] ) is equivalent to the replacement @xmath382 . _ : _ as @xmath532 , @xmath533 disappears on the right - hand side of .
let @xmath383 denote .
then we need to show the following .
@xmath534 for this purpose , multiply the @xmath368-st column of the matrix of @xmath385 with the prefactor @xmath386 . after taking @xmath535 ,
the @xmath368-st column is @xmath536 , and now expand the determinant with respect to this column .
taking @xmath390 in , the last row of the matrix becomes @xmath391 , and we obtain the following simplified formula .
[ dastpartfunct1 ] we have the following formula for @xmath537 .
@xmath538 here we again sketch an alternative proof of corollary [ dastpartfunct1 ] which avoids theorem [ dastpartfunct ] and lemma [ eval ] .
we need to verify @xmath539 . by the symmetry of theorem [ sym ]
, it suffices to show that @xmath540 .
assume @xmath541 and @xmath348 , and consider the bottom bulk vertex @xmath396 in the central column .
its label is @xmath398 , and since @xmath542 , there are only two possible local configurations around @xmath396 that contribute a non - zero weight ( namely @xmath83 and @xmath342 ) .
however , this implies that there are only two possible configurations for the right boundary vertex @xmath403 adjacent to @xmath396 ( the vertical edge connecting @xmath396 and @xmath403 is oriented from @xmath396 to @xmath403 ) .
the assertion follows as @xmath543 .
now the proof is by induction with respect to @xmath29 .
the cases @xmath406 are easy to verify . to show that is a laurent polynomial in @xmath332 , proceed as in the footnote of the proof of theorem [ astpartfunct ] . _ : _ we use the leibniz rule for the determinant .
the expression is a sum of rational functions with degrees and orders of the numerators in @xmath291 being @xmath11 and @xmath544 , respectively , while the degrees and orders of the denominators are @xmath357 and @xmath409 , respectively . _
, , , and : _ the expression is obviously symmetric in @xmath272 , invariant under the transformation @xmath410 and even in @xmath279 , @xmath285 .
the identities from lemmas [ eval1 ] and [ eval2 ] can be verified straightforwardly .
it is also obvious that the expression in has a zero at @xmath350 .
[ dastschurtheo ] we have the following schur function expression for @xmath537 at @xmath238 . @xmath545 use and observe that this expression is a specialization of one of two factors of the partition function of half - turn symmetric @xmath4s as noted in ( * ? ? ?
* theorem 10 ) .
the result follows from ( * ? ? ?
* theorem 2.4 ( 2 ) , 2nd eq . ) or from ( * ? ? ?
* ; * ? ? ?
* eq . ( 11 ) , eq .
( 31 ) ) .
character formulas for the specializations of @xmath546 at @xmath547 are also provided in ( * ? ? ?
* theorem 2.4 ( 2 ) ) .
they correspond to @xmath423-enumerations of @xmath149s for @xmath548 , respectively . in a forthcoming paper @xcite
, we will see that factorizes .
more precisely , it will be shown that @xmath549 we set @xmath550 , @xmath280 , and @xmath551 in to finally prove theorem [ maxone ] .
we use the following formula for the specialization of schur functions .
@xmath552 this is , by the combinatorial interpretation of the schur function as the generating function of semistandard tableaux of shape @xmath553 , also the number of these tableaux .
it follows that @xmath554 and the latter is also the number of cyclically symmetric plane partitions in an @xmath46 box , see @xcite .
a natural question to ask is whether there is a generalization of theorem [ maxone ] that is analogous to theorem [ maxmone ] , i.e. , what is a statistic on @xmath46 @xmath44s that has the same distribution as the numbers of @xmath5 s in the fundamental domain of @xmath21 @xmath1s @xmath15 with @xmath43 ? in a forthcoming paper , we will show that this is the case for a statistic that was already introduced by mills , robbins and rumsey in @xcite , namely the number of special parts in a @xmath44 .
in fact , this result will follow from a common generalization of theorems [ maxmone ] and [ maxone ] that involves an infinite family of alternating sign matrix objects , among which the objects from theorems [ maxmone ] and [ maxone ] are two special members .
in addition , certain classes of extreme diagonally and antidiagonally symmetric alternating sign matrices of _ even order _ will also be members of this family . with regard to this family , it will also make sense to consider objects with vertical symmetry , and we will enumerate these symmetry classes .
a special case of this result will be that vertically symmetric @xmath122s of order @xmath29 are equinumerous with vertically symmetric @xmath4s of order @xmath29 .
more generally , we will see that the generating functions of vertically symmetric @xmath4s and vertically symmetric @xmath122s with respect to the numbers of @xmath5 s coincide .
in this section , we provide the proof of theorem [ maxone ] and generalizations .
recall the definition of the _ pfaffian _ of a triangular array @xmath555 @xmath556 where @xmath557 is the sign of the permutation @xmath558 .
suppose @xmath559 is the skew symmetric extension of @xmath15 .
then it is a well - known fact that @xmath560 [ qoosasmpartfunct ] we have the following pfaffian formula for the @xmath56 partition function @xmath561 where @xmath562 and @xmath563 curiously , @xmath564 is the partition function of @xmath47 @xmath264s using the bulk weights from table [ weights ] and normalizing the boundary weights as in the @xmath56 case ( i.e. , all boundary weights are @xmath6 ) . to see this ,
set @xmath565 in kuperberg s formula ( * ? ? ?
* theorem 10 ) and divide by @xmath566 to account for the different normalization of the bulk weights .
we use theorem [ un ] .
the proof is by induction with respect to @xmath29 .
the case @xmath363 is easy to verify . both @xmath567 and @xmath568 are laurent polynomials as the two pfaffians involved are antisymmetric functions in the variables @xmath569 and @xmath570 , respectively , and odd in each @xmath279 . _ : _ by the definition of the pfaffian
, @xmath567 is a sum of rational functions where the degrees and orders of the numerators in @xmath291 are @xmath571 and @xmath572 , respectively , and of the denominators @xmath573 and @xmath574 , respectively . hence the bounds for the degree and order of @xmath567 are @xmath573 and @xmath574 , respectively .
this shows that , in case of @xmath29 is odd , the degree and order in @xmath229 of the proposed expression for the partition function are at most @xmath64 and at least @xmath575 , respectively ( @xmath576 ) . on the other hand , the degrees ( resp .
orders ) of the numerators of the summands of @xmath568 in @xmath291 are either @xmath571 ( resp .
@xmath572 ) or @xmath577 ( resp .
@xmath578 ) , depending on whether or not @xmath6 is matched to @xmath579 and the degrees ( resp .
orders ) of the denominators are @xmath573 ( resp .
@xmath574 ) and @xmath580 ( resp .
@xmath581 ) .
thus , the degree and order of @xmath568 are at most @xmath573 and at least @xmath574 , respectively .
this shows that , in case of @xmath29 is even , the degree and order in @xmath229 of the proposed expression for the partition function are at most @xmath29 and at least @xmath286 , respectively ( @xmath582 ) . _ , and : _ the expression in the statement is symmetric in @xmath272 as it is the product of two symmetric functions in @xmath272 .
it is also invariant under the replacement @xmath583 as @xmath567 is invariant under @xmath584 and @xmath568 is invariant under @xmath585 , and even in @xmath279 , @xmath285 .
_ : _ let @xmath383 denote the expression in the statement
. then by lemma [ eval ] and by taking into account the normalization we have chosen we need to show @xmath586 thus it suffices to show @xmath587 and this follows easily from the definition of the pfaffian . here
we obtain a formula involving _ symplectic characters _
@xmath588 when specializing at @xmath238 .
( see @xcite for a reference on symplectic characters . )
we use the determinantal formula @xmath589 if @xmath590 , and extend the partition with zero parts if its length is less than @xmath29 .
symplectic characters are also generating functions of certain tableaux : a _ symplectic tableaux _ of shape @xmath553 is a semistandard tableaux of shape @xmath553 with entries @xmath591 and the additional constraint that entries in row @xmath135 are no smaller than @xmath135 , see ( * ? ? ?
* theorem 2.3 ) .
now @xmath592 where the sum is over all symplectic tableaux of shape @xmath553 with entries in @xmath593 .
[ spq ] assume @xmath238 .
we have the following formula for @xmath594 in terms of symplectic characters @xmath595 and for @xmath596 @xmath597 we use theorem [ qoosasmpartfunct ] to prove the theorem .
as @xmath564 is the partition function of @xmath264s , we can take its specialization at @xmath238 from ( * ? ? ?
* theorem 2.5 ( 2 ) ) where it was shown that @xmath598 see also ( * ? ? ?
* theorem 5 8) .
( in ( * ? ? ?
* theorem 2.5 ( 2 ) ) appears also a character formula for the specialization of @xmath564 at @xmath599 . )
we still need to show @xmath600 the pfaffian identity stated next is useful in the following .
[ lem : okadatype ] we define @xmath601 then @xmath602 the following special case of an identity due to okada ( * ? ? ?
* theorem 3.4 .
, eq . ( 20 ) ) is applied .
@xmath603 ( substitute @xmath29 for @xmath64 , and set @xmath604 , @xmath605 , @xmath606 in okada s identity . ) after replacing @xmath607 by @xmath608 in , the right - hand side is up to the factor @xmath609equal to the right - hand side in the statement .
it suffices to show @xmath610 the fact that elementary row and column operations do not change the determinant of a matrix has an analogue for pfaffians : for instance , suppose @xmath15 is an even - order skew - symmetric matrix , and @xmath77 is obtained from @xmath15 by adding a multiple of row @xmath135 to row @xmath235 , and simultaneously adding the same multiple of column @xmath135 to column @xmath235 , for some @xmath611 , then @xmath612 .
the identity can now be proven using such operations .
more specifically , we use the identity @xmath613 for any @xmath614 , which is a special case of a pfaffian analogue of sylvester s determinant identity .
( replace @xmath29 by @xmath64 , and set @xmath615 and @xmath616 in an identity of knuth ( * ? ? ?
* eq . ( 2.5 ) ) .
) alternatively , can be obtained directly by transforming the first @xmath357 entries in the last column of @xmath614 to zeros using certain row and column operations , and then expanding along the last column .
now let @xmath614 be the array in the pfaffian on the left - hand side of .
it can be checked straightforwardly that @xmath617 the result now follows from . we prove .
assume @xmath238 .
as @xmath618 we have that @xmath619 is equal to @xmath620 by setting @xmath621 and @xmath622 in lemma [ lem : okadatype ] , we see that the left - hand side of is , up to the factor @xmath623 ( which arises from a factor @xmath624 in each entry of the triangular array , where @xmath625 for @xmath626 and @xmath627 ) , equal to the pfaffian in the previous expression .
this implies that the previous expression is equal to @xmath628 the required right - hand side of can now be obtained from this expression by multiplying columns @xmath155 of the matrix by @xmath5 , reordering the columns as @xmath629 , simplifying the prefactor , and applying the determinant formula for symplectic characters a different approach to proving theorem [ spq ] would be to show that the function in this theorem fulfills the properties from theorem [ un ] in the special case @xmath238 .
( the same idea would clearly also lead to direct proofs of theorems [ astschurtheo ] and [ dastschurtheo ] , where in this case we would have to show that the properties from theorem [ u1 ] are satisfied .
this approach was previously used , see for instance @xcite . )
it turns out that it suffices to show @xmath630 since @xmath631 is a certain generating function of symplectic tableaux , it would be interesting to explore whether these identities have a combinatorial proof .
we set @xmath550 for @xmath280 in and to compute the number of @xmath56s . here
we need the following product formula , see ( * ? ? ?
* exercise 24.20 ) .
@xmath632 we obtain @xmath633 and this is also the number of @xmath634 @xmath53s . finally , @xmath635 and
this is also the number of @xmath636 @xmath53s , see @xcite .
we thank matja konvalinka for very helpful discussions . the authors acknowledge hospitality and support from the galileo galilei institute , florence , italy , during the programme `` statistical mechanics , integrability and combinatorics '' held in may
june 2015 . the first author ( a.a .
) is partially supported by ugc centre for advanced studies and acknowledges support from dst grant dst / int / swd / vr / p-01/2014 .
the third author ( i.f . ) acknowledges support from the austrian science foundation fwf , start grant y463 .
we count @xmath8 @xmath4s with precisely one @xmath5 : such matrices are uniquely determined by the following information . * the columns of the unique @xmath5 and the two @xmath6 s that are situated in the same row .
there are @xmath637 choices . * the rows of the @xmath5 and the two @xmath6 s that are situated in the same column as the @xmath5 .
there are @xmath637 choices . * the permutation matrix of order @xmath638 that is obtained by deleting the three chosen rows and the three chosen columns .
there are @xmath639 choices .
therefore , there is a total of @xmath640 @xmath4s of order @xmath29 with precisely one @xmath5 .
we note that formulas for the numbers of @xmath4s with more @xmath5 s are given in @xcite and @xcite , and that @xmath4s with one @xmath5 were studied by lalonde @xcite .
on the other hand , @xmath122s with precisely one @xmath5 are less accessible : let @xmath641 be the row of the @xmath5 , counted from the bottom , and @xmath642 its column , counted from the central column and where we use the negative sign if the column is left of the central column .
let @xmath643 encode the positions of the three @xmath6 s to the left , to the right and above the @xmath5 as indicated in figure [ onem1 ] . for integers @xmath644 ,
we use the following notation . @xmath645
we count all possible subarrays consisting of the @xmath646 bottom rows of such an @xmath122 : since there is neither a @xmath6 in column @xmath647 nor in column @xmath648 , the number is @xmath649 ( the argument is analogous to that used for counting @xmath122s without @xmath5 s . )
if we impose the condition that there is also no @xmath6 in column @xmath650 in this subarray then this number is @xmath651 where @xmath652 , @xmath653 , @xmath654 .
there are @xmath655 possibilities for extending the latter type of subarray to a permissible @xmath122 of order @xmath29 . combining and ,
it is clear that there are @xmath656 possibilities for the bottom @xmath646 rows that have a @xmath6 in column @xmath650 . here
, there are @xmath657 ways to extend this to a permissible @xmath122 of order @xmath29 . in total , there are @xmath658 configurations .
it is tedious but straightforward to show that this is indeed equal to @xmath640 .
m. ciucu and c. krattenthaler . a factorization theorem for classical group characters , with applications to plane partitions and rhombus tilings . in _ advances in combinatorial mathematics _ ,
pages 3959 .
springer , berlin , 2009 .
s. sundaram .
tableaux in the representation theory of the classical lie groups . in _
invariant theory and tableaux ( minneapolis , mn , 1988 ) _ , volume 19 of _ i m a vol .
_ , pages 191225 .
springer , new york , 1990 . | for each @xmath0 , we count alternating sign matrices that are invariant under reflections in the diagonal and in the antidiagonal ( @xmath1s ) of fixed odd order with a maximal number of @xmath2 s along the diagonal and the antidiagonal , as well as @xmath1s of fixed odd order with a minimal number of @xmath3 s along the diagonal and the antidiagonal . in these enumerations , we encounter product formulas that have previously appeared in plane partition or alternating sign matrix counting , namely for the number of all alternating sign matrices , the number of cyclically symmetric plane partitions in a given box , and the number of vertically and horizontally symmetric @xmath4s .
we also prove several refinements .
for instance , in the case of @xmath1s with a maximal number of @xmath5 s along the diagonal and the antidiagonal , these considerations lead naturally to the definition of alternating sign triangles which are new objects that are equinumerous with @xmath4s , and we are able to prove a two parameter refinement of this fact , involving the number of @xmath5 s and the inversion number on the @xmath4 side . to prove our results , we extend techniques to deal with triangular six - vertex configurations that have recently successfully been applied to settle robbins conjecture on the number of all @xmath1s of odd order .
importantly , we use a general solution of the reflection equation to prove the symmetry of the partition function in the spectral parameters . in all of our cases ,
we derive determinant or pfaffian formulas for the partition functions , which we then specialize in order to obtain the product formulas for the various classes of extreme odd @xmath1s under consideration . |
following heiles ( 1997 ; h97 ) , we use currently - available observations of the tiny - scale atomic structure ( tsas ) to describe its properties ; the observations have improved , so the inferred properties are now somewhat different . the tsas is seen in 21-cm line absorption so must be part of the cold neutral medium ( cnm ) .
the observations consist of time - variable 21-cm absorption lines against pulsars , vlbi images of the 21-cm line in absorption against extended radio sources , and optical absorption line observations against background stars .
the canonical values for important properties come directly from these observations and have no theoretical or otherwise imaginary input , except for the subjective judgment and weighting that we must perform to collapse a variety of measured values into a single number that we can use as a canonical value .
we provide values for the following properties : 1 .
tsas column density : + @xmath1 + where @xmath2 is the hi column density in units of @xmath3 @xmath4 and @xmath5 the temperature in units of 15 kelvins ( [ equiltemp ] justifies this canonical value of 15 k ) .
the @xmath6-dependence arises because the 21-cm line opacity @xmath7 .
the numerical value 0.75 is 4 times larger than the canonical value adopted by h97 and is based on weisberg & stanimirovic ( this meeting ) .
the tsas scale length in the plane of the sky ( denoted by the subscript @xmath8 ) : + @xmath9 + the @xmath8 arises because with time variability and angular structure , it s the plane - of - the - sky length that s relevant .
3 . the line - of - sight scale length @xmath10 can be different ( by a factor @xmath11 ) because it is unlikely for the structures to be isotropic : + @xmath12 4 .
we combine @xmath13 and @xmath10 to obtain the typical tsas volume density : + @xmath14 5 .
the canonical values for volume density and temperature lead to the pressure @xmath15 : + @xmath16 with the above we make a simple , highly important observation : equation [ peqn1 ] shows that unless the tsas is cold and/or anisotropic , its pressure greatly exceeds not only the typical cnm thermal pressure of @xmath17 @xmath18 k , but also the galactic hydrostatic pressure of @xmath19 @xmath18 k ( boulares & cox 1990 ) .
this single point dominates the rest of the discussion , which includes some theoretical considerations .
is the tsas in thermal equilibrium ?
we normally expect the cnm to attain thermal equilibrium because the time scale is short .
we have ( wolfire et al .
2003 ; wmht ) @xmath20 where the symbols have their usual meanings and @xmath21 is the cooling rate per @xmath18 ; cooling varies as @xmath22 because it is a collisional process . for collisional excitation of cii ( the 158 @xmath23 m line ) by h atoms at @xmath24 k , which is the likely temperature of the tsas , we have @xmath25 or @xmath26 here @xmath27 @xmath18 k , our adopted pressure for the cnm . if the tsas has @xmath28 , then @xmath29 yr .
this is short by most interstellar standards .
however , for the time- and space - variable tsas it might not be short , so we consider two cases , the thermal equilibrium case and the nonequilibrium case . versus volume density @xmath30 .
figure 9 from wmht .
[ wolf9 ] ] suppose first that the tsas is in thermal equilibrium .
then the density and pressure are related as shown in figure [ wolf9 ] , which is figure 9 from wmht . for discussion purposes ,
we use the solid curve for @xmath31 because we envision the tsas to be embedded in cnm . at low and high @xmath30
it defines the warm and cold neutral media with temperatures @xmath32 and @xmath33 k , respectively .
the cnm thermal pressure has a minimum ( @xmath34 @xmath18 k ) located near @xmath35 @xmath18 .
this minimum thermal pressure is defined by microscopic heating and cooling processes , which simply and unequivocally means that the cnm can not exist in thermal equilibrium at lower thermal pressures .
this is a strict lower limit for the cnm thermal pressure in thermal equilibrium .
similarly , the wnm has a microphysical - process thermal pressure maximum ( @xmath36 @xmath18 k ) located near @xmath37 @xmath18 .
there is no such strict microphysical - process upper limit on the cnm thermal pressure .
however , there is an upper limit that arises from pressure - balance considerations .
the cnm occupies a small fraction of the total interstellar volume , so it has to be embedded in warmer gas , either the wnm or the hot ionized medium ( him ) .
if the interface is not dynamic it must have pressure balance .
if the bounding medium is the wnm with no magnetic field , then the cnm thermal pressure must lie below @xmath38 ; otherwise , the wnm is forbidden . however , for dynamical equilibrium it s the _ total _ pressure that counts , not just the thermal pressure .
the wnm probably has the typical interstellar magnetic field of @xmath39 @xmath23 g , which exerts a magnetic pressure of @xmath40 @xmath18 k. this @xmath41 sign is important : it s the magnetic pressure _ difference _ that counts , and even then only for that component that is perpendicular to the field lines , so if the field lines do nt lie along the interface , or if the cnm field is comparable to the wnm field , then the magnetic pressure difference is less .
there s a _ strict _ upper limit to the cnm pressure if it s in thermal and dynamical equilibrium , which is the hydrostatic pressure of the galactic disk ; this is about @xmath42 @xmath18 k ( boulares & cox 1990 ) .
if there s no magnetic field and no turbulence , then the cnm pressure has to lie in the very narrow range @xmath43 to 4810 @xmath18 k , which is a factor of only 2.5 . in equation [ peqn1 ] , for the _ strict _ upper limit
@xmath44 @xmath18 k , we require @xmath45 .
it s unlikely , in our opinion , that the strict upper limit applies always .
a more reasonable value is lower , perhaps @xmath46 @xmath18 k , for which we require @xmath47 .
below we will find that it is impossible to attain @xmath48 k , so this implies @xmath49 .
this seems like a perfectly reasonable degree of anisotropy for tsas , particularly in light of h97 s estimate as @xmath50 for a reasonable choice if tsas is sheetlike . if we do nt have thermal equilibrium all bets are off and we can invoke any conditions we please . from the standpoint of making tsas more easily explained by individual structures , we move toward smallness by invoking higher densities and pressures .
for example , if @xmath51 @xmath18 k , ten times higher than the equilibrium @xmath52 , @xmath53 increases and @xmath54 decreases by a factor of ten .
the decreased cooling time presents the quandary that it s harder to escape thermal equilibrium ! raising the temperature above the lowest equilibrium value of 12
k is counterproductive in the sense of requiring higher hi column densities to produce observable 21-cm line absorption .
raising temperature instead of density also increases the cooling rate even more , because of the highly nonlinear factor @xmath55 in the cooling function @xmath56 ( equation [ ccool ] ; for example , if we increase the temperature from 16 k to a typical cnm temperature of 60 k , the cooling rate @xmath56 would increase by a factor 100 ( ! ) .
conclusion : if we invoke nonequilibrium , it s easier to obtain the observed tsas by keeping it cold and increasing the volume density .
this , of course , also increases the pressure , which leads to some spectacular dynamics in the absence of pressure equilibrium , tiny - scale structures would expand rapidly into the surrounding medium . soon
the structures would no longer be `` tiny '' .
for example , for a pressure ratio between tsas and the surrounding medium of 10 the expansion velocity is close to mach 10 , or @xmath57 km s@xmath58 ; it takes only 500 years to expand by the tsas characteristic scale size of 30 au .
so a structure wo nt last long . on the other hand , one of the ways we detect
tsas is by time variability which is a natural result of dynamical nonequilibrium !
so , in this sense , dynamical nonequilibrium is suggested by the observations .
if the tsas pressure is high , then you have to answer the question : how did the tsas pressure get so high to begin with ?
the answer might be intermittency in the turbulence of the enveloping medium ( hily - brandt , this meeting ; coles , this meeting ) . to summarize : the nonequilibrium case has intrinsically high time variability .
in this sense , it satisfies one of the observational signatures that define tsas !
this poses the question in a slightly different way : the observed time variability shows that tsas is not in perfect dynamical equilibrium , but is it _ far _ from or _ not too far _ from equilibrium ?
the equilibrium temperature is determined by equality of heating and cooling rates . regarding
_ cooling _ : the tsas has no observed molecules , so we are left with the only effective atomic coolant , which is collisional excitation of cii . for cold gas ,
carbon is almost fully ionized and hydrogen almost fully neutral , so @xmath59 is almost equal to the gaseous carbon abundance .
we assume that the c / h abundance is @xmath60 by number and that the _ gaseous _ c / h ratio is @xmath61 times smaller ( some c goes into dust ; people typically assume that the depletion factor @xmath62 ) .
we use wmht equations c1 and c2 to write @xmath63 \ , e^{-92/t } \
\rm{erg \ cm^{-3 } \
s^{-1 } } \ .\ ] ] note that there are two submechanisms : collisions by atomic hi ( the first term ) and by electrons ( the second term ) .
the h - collisions dominate , even there are no grains and @xmath64 . _
heating _ is another matter .
the commonly - accepted dominant heating mechanism is photoionization of pah ; this provides typical cnm temperatures of @xmath65 k. we envision tsas temperatures to be much smaller , so we assume specifically that pah heating ( and , by implication , pahs themselves ) are absent .
this has the ancillary consequence of freeing much of the carbon to the gas phase , making @xmath66 . without pah heating
, we have two possibly important heating mechanisms : photoionization of ci by starlight and ionization of hi by cosmic rays and soft x - rays .
we first consider the former mechanism alone , which provides an absolute lower limit for the temperature of hi regions , and then include the latter . for heating by photoionization of ci ,
the heating rate per unit volume @xmath67 is proportional to the photoionization rate per c atom multiplied by the volume density of c atoms @xmath68 . in ionization equilibrium , the photoionization rate is equal to the recombination rate , and because carbon is almost fully ionized the recombination rate is much easier to calculate .
this gives @xmath69 where @xmath70 is the recombination coefficient .
@xmath71 is the mean energy gained per photoionization ; the uv radiation field drops to zero above 13.6 ev , so @xmath72 ( where @xmath73 is carbon s ionization potential in ev ; see spitzer 1978 , p. 107 , p. 134 - 136 , p. 142 - 143 ) .
remembering that nearly all the electrons are contributed by gaseous carbon , and that carbon is almost fully ionized , we have @xmath74 we get the temperature by equating equations [ ccool ] and [ cheat ]
. if atomic h cooling ( the first term in equation [ ccool ] ) is negligible , then ( amazingly enough ) , as pointed out by spitzer ( 1978 , p. 143 ) , we are left with the ultra - simple equation @xmath75 @xmath76 this is a very robust result : it is independent of both the volume density ( as long as @xmath77 @xmath18 , where collisional de - excitation of cii becomes important ) and the starlight intensity ! now ,
as we remarked above , h cooling does in fact dominate , which lowers the equilibrium temperature to about 12 k. this is a lower limit for the temperature in purely atomic regions .
the neural ism has no uv photons above 13.6 ev energy , so hi can be ionized only by cosmic rays and soft x - rays .
soft x - rays are attenuated by small columns of hi , so their heating rate is sensitive to shielding .
wmht discuss this rather deeply and conclude that a typical primary ionization rate from soft x - rays is @xmath78 s@xmath58 per h atom .
the primary ionization rate from cosmic rays @xmath79 is highly uncertain .
for example , wmht adopt @xmath80 sec@xmath58 per h - atom , a bit larger than @xmath81 .
three years later , tielens ( 2005 ) adopts @xmath82 sec@xmath58 per h - atom , which is @xmath83 ! the difference is at least partly based on the measured abundance of the h@xmath84 ion ( mccall et al 2002 ; le petit , roueff , & herbst 2004 ) .
we have studied a cold ( @xmath85 k ) cloud that ( surprisingly ) lies within the local bubble ( meyer , this meeting and meyer et al .
i had hoped to be able to use the observed 17 k temperature of this cloud to establish an interesting upper limit on @xmath79 within the local bubble .
however , the equilibrium temperature varies slowly with @xmath79 : for @xmath86 , @xmath6 rises to only 14.7 k for @xmath87 . getting @xmath6 as high as 17
k requires @xmath88 .
evidently , the sensitivity of @xmath6 to @xmath79 is small !
just a little pah heating could accomplish the same thing .
common optical interstellar absorption lines , like those of nai , are produced by minority ionization species ( ionization state @xmath89 ) . if the next - higher ionization state ( @xmath90 ) is the dominant one , then we have @xmath91 where @xmath92 is the ionization rate of state @xmath89 and @xmath70 is the recombination coefficient from @xmath90 to @xmath89 . if these two ionization states comprise all of the element , then with @xmath93 the interstellar line strength ( and @xmath94 ) @xmath95 .
typically @xmath96 .
if all the electrons come from carbon , then @xmath97 .
the total @xmath6 dependence is quite steep , @xmath98even steeper than for the 21-cm absorption line of hi [ for which the absorption line strength @xmath99 .
morton ( 1975 ) provides tables of @xmath92 and @xmath100 , where the subscript implies @xmath70 evaluated at @xmath101 k. if one derives @xmath94 at 15 k instead of 56 k , the values are @xmath102 times larger .
more generally , with @xmath93 we have @xmath103 where @xmath104 is the volume density of both ionization states combined . for nai and ci lines , @xmath105 and 11.7 , respectively .
all of this implies that common absorption lines from minority ionization species in the tsas should be strong .
the h@xmath0 molecule is formed on grains and destroyed by ultraviolet photons .
the photon destruction occurs in spectral lines , which allows h@xmath0 to shield itself ; the destruction rate @xmath106 .
tielens ( 2005 ) provides the numbers . for a flat slab illuminated on one side ,
the local volume density h@xmath0 fraction is @xmath107 here @xmath108 is the total number of h - nuclei , including both h - atoms and h - molecules . integrating in from the slab s edge ,
the column density fraction is 4 times smaller .
contrary to h97 , the h@xmath0 abundance is small ; the difference comes from h97 s use of different numerical constants .
this revision is favorable , because now h@xmath0 does nt add to the tsas pressure .
tsas clouds are enveloped by a surrounding medium .
this medium is another gas phase .
possible phases include warmer cnm , wnm , wim , and him .
the interface supports either evaporation of the embedded cloud to the environment , or condensation of the environment onto the embedded cloud .
cowie & mckee ( 1977 ) provided the original treatment of the spherical cloud case .
jon slavin s talk at this meeting provides an excellent summary , including both the essential physics and convenient numerically - expressed equations ; we use his equations in this paper .
these processes are not intuitive , as our short discussion below illustrates . in particular , at the most basic level the mass loss rate depends only on @xmath109 , the thermal conductivity in the external medium , and the cloud radius @xmath110 . for ionized external media , the effective conductivity @xmath109 can be greatly reduced if magnetic field lines lie parallel to the boundary .
contrary to intuition , the mass loss rate per unit area is independent of cloud volume density @xmath111 .
a very important concept is heat - flux saturation , which produces an upper limit on the evaporation rate . for unsaturated evaporation ,
the most important and surprising point is that the mass loss rate _ per unit area _ of cloud is _ not _ independent of the radius of curvature @xmath110 , but rather is proportional to @xmath112 ; this results from the scaling of the gradient for spherical geometry .
this means that , for a spherical cloud , the overall evaporation rate @xmath113 instead of @xmath114 and the evaporation lifetime @xmath115 instead of @xmath110 .
in contrast , the saturated evaporation rate per unit area @xmath116 so the lifetime @xmath117 .
a second important concept is evaporation versus condensation : small clouds lose mass by evaporation , while large clouds gain mass by condensation .
the critical radius @xmath118 is given by cowie & mckee figure 2 .
for cnm embedded in wnm at @xmath119 @xmath18 k , @xmath120 au . if tsas clouds are spherical , then they are smaller than this so they should evaporate .
similarly , tsas should also evaporate when embedded in him .
but tsas is almost certainly not spherical !
the shapes of clouds really matter for evaporation .
thus , flat clouds , whose faces have infinite @xmath110 , never evaporate but instead grow by condensation on their faces . however , their edges can be sharp with small radii , where they can evaporate .
evaporation at the edges and condensation on the faces tends to push them towards sphericity .
tsas evaporation is unsaturated when embedded in wnm and saturated in him .
the evaporation timescale for spherical clouds having radius equal to the canonical @xmath121 au of equation [ tsaslperp ] is about 500 yr ( this time @xmath115 ) and 5000 yr ( this time @xmath122 ) in the wnm and him , respectively .
these are short timescales , comparable to the dynamical timescale we estimated in [ dynamics ] both times are much shorter than times for thermal equilibrium
. there might not be a sharp interface .
for example , tsas clouds are probably very cold and have no grain heating .
they might be surrounded by a thicker cnm envelope that contains grains , in which the temperature gradually increases outwards to perhaps 100 k or more ; this would be an effective blanket against tsas evaporation .
section [ one ] provides canonical values for tsas , assuming that it consists of discrete structures .
section [ two ] compares equilibrium versus nonequilibrium . in _
thermal _ equilibrium , with rough pressure equality the tsas is so cold that the cooling rates are small ; the thermal equilibrium time scale is @xmath123 yr , which might be too long for thermal equilibrium to be established .
if the tsas is not in thermal equilibrium , then it is probably warmer than 15 k , which exacerbates the tsas overpressure problem and leads to _ dynamical non_equilibrium .
if the tsas is not in _ dynamical _ equilibrium , meaning that its internal pressure exceeds that of its surroundings , then the overpressure makes tsas clouds expand , probably forcibly .
tsas exhibits time variability , which is not inconsistent with such expansion .
however , if the tsas is far from equilibrium , one must concoct a mechanism for its formation .
section [ three ] discusses some physical processes in tsas .
tsas is probably very cold , near the lower limit of @xmath124 k for optically thin molecule - free gas .
optical lines of minority species get stronger at low temperatures , and these lines should be strong in the tsas .
evaporation and/or dynamical expansion limits tsas lifetimes for spherical clouds , but maybe not for anisotropic clouds , magnetic fields , or unsharp cloud boundaries .
molecular hydrogen should not be plentiful in tsas . | almost a decade ago i wrote an article with the same title as this .
it focused on the physical properties of the tiny - scale atomic structure ( tsas ) as discrete structures of the cold neutral medium ( cnm ) . to be observable , tiny discrete structures that do nt grossly violate pressure equilibrium need two attributes : low temperatures and geometrical anisotropy . here
i update that article .
i discuss thermal and pressure equilibrium , ionization , optical lines , h@xmath0 abundance , and evaporation . |
classical gamma - ray burst ( grb ) afterglow models generally invoke an `` impulsively '' injected ( possibly collimated ) fireball running into an ambient interstellar medium ( ism ) or into a pre - burst stellar wind . in principle , after the initial fireball starts to decelerate , additional injection of energy is still possible .
there are at least three motivations to study the post - burst injection possibility more carefully . 1 : additional injection is a natural expectation of the internal shock grb model , which predicts that some slow shells will trail the outermost fast shell and run into it when the latter is decelerated ( e.g. rees & 1998 ; kumar & piran 2000 ) . in some cases ,
the central engine may also inject fast shells at a later time .
later injection is also naturally expected in certain types of the central engine models ( e.g. dai & lu 1998 ; zhang & 2001a ) .
2 : a grb fireball can in principle be dominated by the kinetic energy of the baryons or by a poynting - flux component .
any clue about the fireball nature is presently lacking . as discussed below , an injection signature in the grb lightcurve
may directly provide diagnostics about the nature of the injection as well as information about the central engine .
3 : `` bump '' features have been observed in several grb afterglows ( e.g. grb 970228 , grb 970508 , grb 980326 , grb 000103c , etc . ) , and various interpretations have been proposed , e.g. , refreshed shocks ( panaitescu , & rees 1998 ) , supernova components ( bloom et al . 1999 ; reichart 1999 ; galama et al .
2000 ) , dust echos ( esin & blandford 2000 ) , and gravitational micro - lensing events ( garnavich , loeb & stanek 2000 ) .
it would be of great interest to study the injection feature in more detail to find distinct properties to be differentiated from other interpretations .
such injections are likelier to occur in the early afterglow phase . the planned future broad - band grb mission _
swift _ will have the ability of recording broadband early afterglow signals from many grbs , and thus bring unique opportunities to study the injection features .
depending on the different types of the central engine , the post - injection could consist of , e.g. , some kinetic - energy ( i.e. baryon ) dominated shells , or a poynting - flux - dominated wind ( usov 1994 ; & rees 1997b ) . here
we present a generic study on various injection cases . generally speaking
, an injection energy comparable to the initial impulsive energy is required to make the injection noticeable .
we therefore study a system containing an impulsive shell which is already heated during the shell - ism interaction , and which is collecting material from the ism , and in the meantime also receives a large enough injection energy from a continuous poynting wind or a trailing matter shell .
we first discuss in 2 the simplest case , in which the fireball continuously receives pure energy with negligible baryon loading , very likely in the form of poynting flux .
since no reverse shock is expected in such a case , the injection signature is solely from the forward shock emission , through the change of the blastwave global dynamics . in 3 , we discuss the more complicated matter shell injection case . in real situations , shell collisions may occur more than once . here
we only investigate the detailed physics of one such post - collision .
the same analysis , when applied several times , is generic in delineating the more complicated multi - collision cases .
we first analyze the general hydrodynamics of the three - shell ( impulsive shell , injective shell , and ism ) interaction ( 3.1 ) . after reviewing the general shell evolution history
, we solve numerically for the condition of a `` violent injection '' , in which an additional pair of strong shocks form at the discontinuity of the two colliding shells ( 3.2 ) .
we then outline in 3.3 the hydrodynamics of the three - shell - interaction process in detail by dividing the whole process into five stages ( 3.3.1-3.3.5 ) , namely ( 1 ) pre - collision ; ( 2 ) mild injection ; ( 3 ) violent injection ; ( 4 ) relaxation ; ( 5 ) post - relaxation . in 3.4 , we discuss synchrotron radiation from all the possible shocked regions in various stages , and present the injection signatures in the broadband lightcurves .
our findings are summarized and discussed in 4 .
the conventional definition of a poynting - flux - dominated flow is @xmath0 , where @xmath1 is the luminosity of the electron - position pair , radiation and nucleon components , and @xmath2 is the luminosity of the poynting - flux ( i.e. magnetic field ) dominated outflow component ( e.g. usov 1999 ) . in the context of grbs ,
discussions of a poynting - dominated flow as compared to a kinetic - energy - dominated flow are mainly focused on the paradigm of an approximately baryon - free poynting flow , which are not as conducive to collisionless shock formation , so in this case grbs may be powered by the non - linear breakdown of large - amplitude electromagnetic waves ( usov 1994 ; blackman & yi 1998 ; lyutikov & blackman 2001 ; for a review , see usov 1999 ) .
however , in both scenarios , collisionless external forward shocks are believed to form in the afterglow phase ( usov 1999 ) . although the baryon free nature at the grb prompt phase is uncertain ( e.g. spruit , daigne & drenkhahn 2001 ) , in the later injection phase the bayron loading could be in principle much lower and the injection could be in the form of an almost bayron - free poynting - flux - dominated wind .
in such a case , no collisionless reverse shock is expected to propagate into the wind ( kennel & coroniti 1984 , usov 1999 ; essentially , because the sound speed in the ejecta is close to the speed of light ) , and the injection energy is used to increase the total energetics of the fireball . considering an adiabatic relativistic hot shell which is collecting material as it expands into the ism and which receives an energy input from the central engine via a poynting - flux - dominated flow , the differential energy conservation equation in a fixed frame is @xmath3=d m_2c^2 + d e_{\rm inj}$ ] , where @xmath4 , and @xmath5 are the masses of the impulsive shell and the swept - up ism , respectively ) . ] , @xmath6 is the received injection energy , and @xmath7 is the bulk lorentz factor of the blastwave . the injection wind can be approximated as being cold , so it does not provide any internal energy , and the total internal energy in the fireball is provided by the shock heated ism .
noticing equation ( [ 21 ] ) , one can usually approximate @xmath8 as a working assumption to perform a first - order treatment of the problem . strictly speaking ,
such an approximation is only valid when the post - shock fluids are considered uniform , i.e. , with constant density , temperature , and lorentz factor .
more detailed self - consistent modelings require taking into account the blandford - mckee ( 1976 ) self - similar profiles and tracking the post - shock internal energy more closely . with the assumption
, the integrated energy conservation equation then reads @xmath9 this is analogous to the combination of equations ( 1 ) and ( 2 ) in zhang & ( 2001a ) .
when @xmath6 is negligible and noticing that @xmath10 , equation ( [ e ] ) is the standard equation for the impulsive adiabatic blastwave evolution ( e.g. chiang & dermer 1999 ; huang , dai & lu 1999 ) , which can be reduced to the familiar form @xmath11 ( so that @xmath12 , where @xmath13 is the blastwave radius in the fixed frame , and @xmath14 is the observer time ) when @xmath15 . in the presence of @xmath6 , we can see that the condition for changing the blastwave dynamics is that @xmath6 exceed @xmath16 , or essentially @xmath17 .
a poynting - flux - dominated wind is usually continuous , with an intrinsic luminosity law which may be defined through a temporal index @xmath18 within a certain time regime , e.g. @xmath19 , where @xmath14 is the intrinsic time of the central engine ( which is the same as the observer s time after the cosmological time dilation correction ) . a detailed description of such an injection scenario , especially in the context of magnetar central engine models ,
was discussed in a previous paper ( zhang & 2001a ) . here
we summarize some of the conclusions .
when the blastwave dynamics is dominated by the injection ( i.e. @xmath20 , which only occurs when @xmath21 ) , the bulk lorentz factor evolves according to @xmath22 . in most problems ,
usually @xmath23 during the injection phase , so that @xmath24 .
there are three relevant time scales in the problem , i.e. , the blastwave deceleration time , @xmath25 ; the critical time when the continuous - injection energy component dominates the impulsive - injection energy component , @xmath26 ; and the characteristic time durung which the central engine produces a substantial energy imput ( i.e. @xmath21 ) , @xmath27 .
the condition for the injection signature to show up in the afterglow lightcurves is @xmath28 . for @xmath29
, the dynamics is defined by the continuous - injection as soon as the afterglow is set up , and we define it as an immediate injection ( fig.1a ) .
the lightcurves will be flat from @xmath25 to @xmath27 , and will steepen after @xmath27 . for @xmath30 , the signature will not show up until @xmath26 , and we define it as a delayed injection ( fig.1b ) .
there are two temporal index changes : the lightcurves will flatten around @xmath26 , and resume the original steepness around @xmath27 .
the injection - signature is achromatic due to the change of the blastwave dynamics .
since no reverse shock is expected , only emission from the forward shock is responsible for the afterglow emission .
assuming synchrotron emission of the relativistic electrons , the temporal index at the injection phase is @xmath31 ( which is @xmath32 for @xmath23 ) , as compared to the standard case of @xmath33 , where @xmath34 is the spectral index in a certain band of interest .
as an example , we again discuss the poynting - flux injection case in the magnetar model . in this particular case , the injection luminosity law is @xmath35 , where @xmath36 and @xmath37 , and @xmath38 , @xmath39 are the pulsar dipolar field strength at the pole , and the initial period at birth , @xmath40 and @xmath41 are the rotation inertia and the radius of the pulsar , respectively .
figure 1 shows two example injection lightcurves for both the immediate and the delayed injection cases in three energy bands .
a general feature is that the injection signature is rather smooth , mainly due to the smooth varying luminosity law @xmath42 .
dai & lu ( 2001 ) and wang & dai ( 2001 ) obtained a similar result recently .
the previous poynting - dominated injection is , for the present purpose , the simplest case , in the sense that the injection does not result in new emission sites .
the signature in the lightcurves is still produced by the shocked ism , albeit through a changed global dynamics of the blastwave . however , if the injection energy is dominated by the kinetic energy of the baryons , collisionless shocks may in principle form at the discontinuity of any two adjacent shells under certain conditions .
the emission sites during the injection process therefore could be multiple , and the injection signatures , due to the joint contributions from all these sites , should be much more complicated . in this section ,
we investigate in some detail the shell collision process , as well as the emission signatures .
we consider a collision in the afterglow phase , in which case the impulsive shell is already heated up by the passage of the reverse shock . assuming that the electron equipartition factor @xmath43 is not close to unity
, the impulsive shell will not cool at the time of injection , even if the electrons cool rapidly .
a generic configuration of the problem is summarized in figure 2 , which is predicated upon the fulfillment of certain conditions ( 3.2 ) .
there are six regions of interest , namely : ( 1 ) unshocked ism ; ( 2 ) forward - shocked ism ; ( 3 ) unshocked hot impulsive ( leading ) shell ; ( 4 ) forward - shocked impulsive shell ; ( 5 ) reverse - shocked injective ( trailing ) shell ; ( 6 ) unshocked injective shell .
we denote through @xmath44 , @xmath45 , @xmath46 and @xmath47 the mass , baryon number density , internal energy density , and the pressure of the region `` @xmath48 '' in its own rest frame , respectively ; @xmath49 and @xmath50 are the bulk lorentz factor and the dimensionless velocity of region `` @xmath48 '' measured in the fixed frame , respectively ; and _ ij ( _ i/_j+_j/_i)/2 , for _ i , j 1 [ gamij ] and @xmath51 the bulk lorentz factor and the dimensionless velocity of region `` @xmath48 '' measured in the rest frame of region `` @xmath52 '' , respectively . in the hot regions ( 2 , 3 , 4 and 5 ) , we assume a relativistic equation - of - state , i.e. , @xmath53 , where @xmath54 .
two contact discontinuities separate the three parts , i.e. , the ism , the impulsive shell ( which we will consistently call shell 3 in the following discussions ) , and the injective shell ( which we call shell 6 in the following discussions ) .
the pressure equilibrium and velocity continuity at the contact discontinuities yield e_3=e_2 , & _ 3=_2 , [ e32 ] + e_5=e_4 , & _ 5=_4 .
[ e54 ] there are three shocks that accelerate particles and contribute to the emission , i.e. , ( a ) a first forward shock propagating into the ism , ( b ) a second forward shock propagating into the impulsive shell , and ( c ) a reverse shock propagating into the injective shell .
the continuity of the energy , momentum , and particle flux densities in the rest frame of each shock leads to the following jump conditions ( blandford & mckee 1976 ; sari & piran 1995 ; kumar & piran 2000 ) = ( _ 2 - 1 ) m_pc^2 , = 4_2 + 3 , [ 21 ] _ 43 ^ 2= , [ gam43 ] ( n_4 n_3)^2= , [ n43 ] = ( _ 56 - 1 ) m_pc^2 , = 4_56 + 3 . [ 56 ] notice that the jump conditions at the second forward shock ( eqs .
[ [ gam43 ] ] and [ [ n43 ] ] ) are different from the others , mainly because the region in front of the shock ( region 3 ) is hot . with ( [ e32 ] ) and ( [ e54 ] ) , a combination of ( [ 21 ] ) and ( [ 56 ] ) yields = ( 1 , ) , [ e4/e3 ] where f()=(-1)(4 + 3 ) 4 ^ 2 , 1 .
[ f ] the second half of the equation ( [ e4/e3 ] ) will be explained further below . given @xmath55 and @xmath56 , the unknown @xmath57 can be solved for by replacing ( [ e4/e3 ] ) into ( [ gam43 ] ) with the use of ( [ gamij ] ) .
equations ( [ 21])-([56 ] ) imply that strong shocks form between the contact discontinuity of the injective shell and the impulsive shell ( e.g. blandford & mckee 1976 ) , as indicated in figure 2 .
a physical solution for the non - linear equations ( [ 21])-([56 ] ) requires @xmath58 , or @xmath59 , which demands a minimum relative lorentz factor between the two colliding shells given the parameters of both shells ( eq.[[e4/e3 ] ] ) .
this is qualitatively in agreement with kumar & piran ( 2000 ) , who demand a minimum enthalpy density ratio for a given @xmath60 ( in their fig.3 ) , which is equivalent to our case in which we demand a minimum @xmath60 for a given @xmath61 .
for a shock to form , the injective shell should move super - sonically with respect to the impulsive shell , i.e. , by noticing the sound speed @xmath62 for the shell 3 ( in the relativistic limit @xmath63 ) , @xmath64 .
usually this condition is automatically satisfied given the strong shock forming condition @xmath65 .
in this subsection , we examine the criterion for @xmath65 , which is the condition for the configuration in figure 2 being realized .
we first summarize the evolution of a matter fireball shell ( e.g. , laguna & rees 1993 ; piran , shemi & narayan 1993 ; kobayashi , piran & sari 1999 ; piran 1999 ) , and then show how a shell collision process is well - defined by the initial conditions of the shells .
the injection of a matter shell from the central engine can be quantified by three independent parameters , i.e. , the energy of the shell @xmath66 , the mass of the shell @xmath67 ( thus the dimensionless entropy , or the maximum lorentz factor of the shell is @xmath68 ) , and the initial thickness of the shell @xmath69 which is essentially determined by the timescale of the central engine activity , i.e. , @xmath70 .
the following parameters can be consequently defined : the initial baryon number density of the shell , which is @xmath71 $ ] ; the radius at which the fireball stops accelerating and enters the coasting phase , which is @xmath72 ; and the radius at which the fireball starts to spread with its local sound speed due to the velocity difference within the fireball matters , which is @xmath73 .
there is another parameter , i.e. , the deceleration radius @xmath74 , which usually appears in the problem .
however , this radius not only depends on the intrinsic parameters of the shell ( e.g. total energy ) , but also on the properties ( e.g. density and velocity ) of the medium that stops the shell .
for the interaction between the impulsive shell ( shell 3 ) and the ism , the deceleration radius is roughly the radius where the mass of the ism collected by the fireball is @xmath75 of the shell mass , i.e. , @xmath76 , where @xmath77 is the ism baryon number density in unit of 1@xmath78 , @xmath79 ( hereafter @xmath80 , except those subscripts that denote different regions , such as @xmath77 ) .
for the interaction between the injective shell ( shell 6 ) and the impulsive shell ( shell 3 ) , the deceleration radius is around the radius where the injective shell catches up with the impulsive shell .
let us denote @xmath81 as the time interval between ejecting the two adjacent shells ( as compared to @xmath82 , which is the timescale of the injection process ) , the `` catching - up '' radius is then @xmath83 . for @xmath84 ,
@xmath85 is mainly defined by the lorentz factor of the slow , leading shell , i.e. , @xmath86 .
the condition for the collision to occur in the afterglow phase is then @xmath87 , or approximately @xmath88 .
the deceleration radius of a shell can in principle be either smaller or larger than the spreading radius , @xmath89 . for @xmath90 ( a `` thick shell '' , e.g. sari & piran 1995 ) ,
the shell has not started to spread when the deceleration ( or the collision ) occurs .
alternatively , for a `` thin shell '' ( with @xmath91 ) , the shell is spreading with its sound speed in the comoving frame due to the velocity difference within the shell as the collision occurs . in this sense , the initial `` thin shell '' is actually thick during the collision .
table 1 summarizes the evolution of the shell widths ( both in the fixed frame , @xmath92 , and in the comoving frame , @xmath93 ) , shell baryon number density @xmath94 , and the ratio @xmath95 in various regimes . in the last column ( i.e. , @xmath95 ) , for the rear shell 6
, we have assumed that the radius of the shell 6 front is comparable to that of the shell 3 front , i.e. , we assume that @xmath96 , which is generally the case as long as @xmath97 . we note
that expressing @xmath95 in terms of the mass ( or energy ) ratio of the shell and the collected ism ( @xmath5 ) allows a clearer physical understanding for the problem . at the collsion point @xmath85 , it is convenient to show that @xmath98 ( thick shells for the former , and thin shells for the latter ) . noticing that @xmath99 , and that for @xmath100 ( i.e. , the collision occurs long after the afterglow s setting - up but well before the non - relativistic phase , which is usually the case ) , one has @xmath101 , and @xmath102 ( see eq.[[stg:2 ] ] and the discussions thereafter ) .
this justifies the second part of the equation ( [ e4/e3 ] ) .
the condition for @xmath65 can be investigated numerically .
figure 3 shows the minimum @xmath60 required for @xmath65 for different energy ratios of the two shells , @xmath61 .
notice that the wiggles on the curves are due to the crudeness of the numerical solutions .
a general trend is that a larger relative velocity between the two shells is required for a less energetic trailing shell .
the critical line depends on the status of the shell 6 , i.e. , whether it is expanding ( thin shell ) or not ( thick shell ) when the collision occurs .
this is least demanding for an expanding ( thin ) shell 6 , which is plotted as the solid curve . for a thick shell 6 , the larger the ratio @xmath103 ,
the more demanding the condition is .
this is because for the same energy ratio @xmath61 , a thicker shell tends to get a smaller @xmath104 ratio ( eq.[[e4/e3 ] ] ) .
the dotted and the dashed curves in figure 3 are for a shell with @xmath105 5 and 10 times the collision radius , @xmath85 , respectively .
another comment is that , as long as in the relativistic phase lasts , only the energy ratio and the relative velocity are relevant to the problem .
the absolute values of the shell energies and lorentz factors do not enter the solution .
notice that the supersonic condition defines another critical line for the shock forming condition ( fig.3 ) .
the case for @xmath106 not being satisfied is less clear , but in this case no strong shock is expected .
we regard such an injection case as `` mild '' ( with weak or non - shock ) , and we expect that the injective materials are effectively attached onto the forward shell to increase the total energy budget , which is analogous to the poynting - flux injection case . in summary ,
the input parameters in the injection problem include : ( i ) @xmath107 ; ( ii ) @xmath108 ( or @xmath109 , which is only important around the on - set of the afterglow ) ; ( iii ) @xmath77 ; ( iv ) @xmath110 ; ( v ) @xmath111 ( or @xmath112 ) ; ( vi ) @xmath113 , which determines @xmath114 ; ( vii ) @xmath81 ( or equivalently @xmath85 ) .
the comparison between @xmath105 and @xmath85 then decides whether the trailing shell is `` thin '' or `` thick '' when the collision occurs .
given a set of injection parameters ( i ) - ( vii ) , at any radius @xmath13 , one can first check the shock forming condition in figure 3 .
if the condition is not satisfied ( below the critical line ) , the injection is a `` mild '' one .
if the condition is satisfied ( above the critical line in fig.3 ) , the shell - merging process is violent . though an injection of matter shell could be mild or violent throughout , in some cases the injection could be mild in the beginning , and turn violent at larger radii when @xmath55 drops so that @xmath60 meets the critical value . in this subsection
, we present an example which includes both cases during the injection .
there are four characteristic radii in the problem : ( 1 ) the collision radius , @xmath115 ; ( 2 ) the radius where the persistent shocks form ( fig.3 ) , @xmath116 ; ( 3 ) the radius where the injection process ceases , @xmath117 ; ( for a mild injection , this is the radius where all the material in the shell 6 catches up with and `` attaches '' onto shell 3 ; for a violent injection , this is where the reverse shock crosses the rear shell and the merging process enters the relaxation stage . )
( 4 ) the radius where @xmath104 drops to unity and the relaxation finishes , @xmath118 .
the whole shell - merging process may be then be ( at most ) divided into five stages ( fig.4a ) : ( 1 ) pre - collision , @xmath119 ; ( 2 ) mild injection , @xmath120 ; ( 3 ) violent injection , @xmath121 ; ( 4 ) relaxation , @xmath122 ; and ( 5 ) post - relaxation , @xmath123 .
below we discuss each stage in some detail .
during this mild injection phase , the mass in the shell 3 gradually increases while that in the shell 6 ( un - collided part ) gradually decreases .
this ends when the @xmath65 condition is satisfied at @xmath116 and a pair of shocks ( b and c ) propagate into the two colliding shells , or , in a less energetic situation , when the whole shell 6 is `` attached '' onto the shell 3 and the injection process ceases . for the former case ,
we define @xmath136 and @xmath137 , both of which will be used later in the stage 4 ( 3.3.4 ) .
if the shell 6 is energetic enough or fast enough ( see fig.3 ) , one then has the configuration shown in figure 2 .
the two new shocks ( b and c ) propagate into the two colliding shells , respectively . in principle , one can define this stage to end when either of the two shocks crosses the corresponding shell .
however , the reverse shock always crosses the rear shell first , and we define the radius where this happens as @xmath117 .
the violent injection stage is then defined as @xmath138 . to describe the physical process at this stage
, we need to quantify the dynamical evolution of the three shocks , or equivalently , the temporal evolution of @xmath139 and @xmath140 . strictly speaking , @xmath55 and @xmath57
should be solved self - consistently , and one needs one more independent relation between @xmath55 and @xmath57 besides ( [ gam43 ] ) and ( [ e4/e3 ] ) , in order to get a self - consistent solution . however
, this is usually not necessary at this stage .
again , we write down the global energy conservation of the whole system , i.e. , @xmath141 . noticing @xmath142 , @xmath143 , this gives a time - dependent equation ( _ 0,6-_4_56)m_5-(_4-_3)m_4 + + ( _ 0,3 -_3)m_0,3=(_3 ^ 2 - 1)m_2 . [ stg:3 ]
comparing to equations ( [ stg:2 ] ) and ( [ e ] ) , we find that the first term in those equations ( the already injected energy ) is now replaced by the first two terms in ( [ stg:3 ] ) .
the global dynamics ( @xmath55 evolution ) is modified only when the sum of the first two terms is comparable to the third term which is essentially @xmath144 .
this is impossible , since the energy already injected into the system ( i.e. @xmath145 ) is now partially consumed in maintaining the kinetic energy of the injected material [ i.e. , @xmath146 , since for @xmath147 the term @xmath148 is nearly zero ] , and in raising the lorentz factor of the newly shocked regions [ i.e. the term @xmath149 .
unless @xmath150 , which is rather unlikely in grbs , the sum of the first two terms is negligibly small , and ( [ stg:3 ] ) is essentially reduced to ( [ stg:1 ] ) .
the simple scaling law of @xmath55 for the impulsive injection case is then applicable at this stage , e.g. , for an isotropic adiabatic fireball running into a uniform ism , one has @xmath151 .
the unknown quantity @xmath57 , of interest here , can then be solved for by using ( [ gam43 ] ) and ( [ e4/e3 ] ) .
the temporal evolution of both @xmath152 and @xmath153 , as well as the random lorentz factor in the region 4 , @xmath154 , are thus well quantified .
although generally this is done numerically , a rough analytic estimate is possible .
when the solution of @xmath57 is physical , one can approximate @xmath155 , as long as @xmath156 . in this limit
, @xmath157 , @xmath158 , and one has @xmath159 . for our assumption of @xmath160const
, we have @xmath161 for a spreading shell ( @xmath162 for a non - spreading shell )
. we can further estimate @xmath163 ( or @xmath164 ) . since @xmath165 , we have @xmath166 ( or @xmath167 ) . with ( [ n43 ] )
, we also have @xmath168 ( or @xmath169 ) . noticing that the shell 3 is spreading and that @xmath170 , one then has @xmath171 ( or @xmath172 ) .
the `` random '' lorentz factor in the region 4 ( at the shock b ) which defines the radiation from the region ( notice that it is different from @xmath173 ) is then @xmath174 ( or @xmath175 ) . finally , the random lorentz factor at the shock c is essentially @xmath176 , which scales as @xmath177 ( or @xmath178 ) .
these very crude estimates generally agree with the numerical results ( see fig.4a ) .
finally we estimate @xmath117 .
a self - consistent calculation of the shock crossing time ( for both shells ) requires hydrodynamically tracking the speeds of the shocked and un - shocked shells .
very roughly , the shock crossing time ( e.g. , for the rear shell 6 ) may be estimated ( sari & piran 1995 ) @xmath179 \cdot ( 1-{\gamma_{6 } \bar n_6}/ { \bar\gamma_5 \bar n_5 } ) \sim ( \delta_6/c ) \bar\gamma_4 ^ 2 $ ] ( for @xmath180 ) , where @xmath181 denotes the average value of the quantity @xmath182 during the passing process .
this estimation is only good for a not very long passing time , so that the variables do not change much during the crossing period .
this is the case for the rear shell , but not for the forward shell .
for example , for a thin rear shell , @xmath183 , and @xmath184 .
this is much less than the dynamical time of the shell propagation in the fixed frame , @xmath185 , as long as @xmath186 . for a thick rear shell ,
the crossing time is even shorter since the shell is not spreading . for the forward shell , however , one has @xmath187 , which is very long .
therefore it is evident that @xmath188 is always shorter than @xmath189 , and we use it to define the end of the collision stage . the radius @xmath117 is then defined as @xmath190 , and we denote the corresponding parameters at this radius with a subscript `` @xmath92 '' , e.g. , @xmath191 , @xmath192 , @xmath193 , @xmath194 , etc .
these quantities are recorded from the code and are inputs to the dynamical problem of the next stage ( see below ) .
beyond @xmath117 , the reverse shock passes across the rear shell , and the process enters the dynamical relaxation stage .
after this point , there is no further significant energy input into the region ( 4 + 5 ) . since @xmath65 is still satisfied at @xmath117 , the forward shock b still exists and propagates into the forward shell as long as the region ( 4 + 5 )
is separated from the region ( 2 + 3 ) super - sonically .
the region ( 4 + 5 ) then expands adiabatically , since there is practically no heat flow across the shock . the internal energy density in the region ( 4 + 5 ) decreases according to the adiabatic law , i.e. , @xmath195 for @xmath196 .
as a result , @xmath104 as well as @xmath173 drops rapidly towards unity . during this process
, @xmath55 increases and @xmath57 drops , until the whole region ( 2 + 3 + 4 + 5 ) reaches the same velocity , and contacts sub - sonically .
the shock ( b ) dissappears and the relaxation stage ends , at the radius defined by @xmath118 . during the relaxation stage , the shock conditions ( [ 21 ] ) , ( [ gam43 ] ) and ( [ n43 ] ) still hold .
again one needs to quantify the dynamical evolution of @xmath55 and @xmath57 . unlike the previous stage , @xmath55 no longer evolves according to the simple @xmath197 form .
let us again write down the global energy conservation of the whole system , i.e. , @xmath198 .
noticing that the shock heating energy produced at ( a ) is only consumed in the region ( 2 + 3 ) , i.e. , @xmath199 , this gives ( _ 0,6-_4)m_0,6-(_4-_3)m_4 -_4 u_4 + 5 + + ( _ 0,3-_3 ) m_0,3=(_3 ^ 2 - 1)m_2 .
[ stg:4 ] this is even more complicated than ( [ stg:3 ] ) .
as the relaxation goes on , @xmath57 gets closer to @xmath55 and @xmath200 drops rapidly , thus both the second and the third terms decline . the first term , which is essentially the injected energy , starts to influence the blastwave dynamics , just like the poynting - flux injection or the mild injection ( stage 2 ) cases ( cf .
eqs.[[e ] ] and [ [ stg:2 ] ] ) .
the relaxation is a complicated process .
strictly speaking , one needs to solve ( [ stg:4 ] ) and ( [ gam43 ] ) to find a self - consistent solution for @xmath55 and @xmath57 .
here we perform a simpler but adequate treatment .
since @xmath55 is boosted up relative to the non - injection case , we assume that _ 3 ^ 2 e_3 r^-3 + , [ delta ] where @xmath201 is an unknown positive index . at @xmath118 ,
the internal energy density of the whole fireball is then @xmath202 . since beyond @xmath118 the fireball evolves as if the initial total energy is @xmath203 , then @xmath204 can be also expressed in an adiabatic expansion form , but as if the energy density at @xmath117 is a factor @xmath205 times larger , i.e. , @xmath206 .
this gives = .
the last task is to determine @xmath118 . in the relaxation stage
, @xmath207 can be expressed as @xmath208 , according to the assumption made . on the other hand
, @xmath209 evolves according to the adiabatic law , i.e. , @xmath210 .
one then has @xmath211 . at @xmath212 , @xmath213 and @xmath214 by definition .
noticing again equation ( [ delta ] ) , the relaxation stage ends at r_f = r_()^1/3 ( ) ^4/9 ( 1+)^-1/9 .
[ rf ] since the parameters at @xmath117 are in principle known ( 3.3.3 ) , equations ( [ delta])-([rf ] ) , together with ( [ 21 ] ) and ( [ gam43 ] ) , adequately describe the dynamics in the relaxation stage .
some crude estimates match the numerical results . with @xmath215 , @xmath216 and @xmath217 ( eq.[[gam43 ] ] ) , one can get , roughly , @xmath218 , a very steep declination ( fig.4a ) . as long as the shock ( b ) still exists , new electrons are accelerated , and strong emission from the region heated by the shock ( b ) continues throughout the relaxation stage . at the rear end of the shell ( the region 5 ) , since there is no longer a reverse shock ( c ) , no new electrons are accelerated .
however , the already accelerated electrons are still hot and radiate synchrotron emission .
the random lorentz factor in the region 5 , which we still denote as @xmath219 ( although the shock ( c ) no longer exist ) , also drops rapidly . since @xmath220 , and @xmath221 ,
one has @xmath222 ( fig.4a ) .
since @xmath104 no longer @xmath223 , there is no simple scaling - law estimation of @xmath154 at this stage .
after the relaxation ends , one has @xmath224 , the original @xmath200 has died out , and the new thermal energy produced at the shock ( a ) is consumed in the whole shell ( 2 + 3 + 4 + 5 ) , i.e. , @xmath225 .
the global energy conservation now reads @xmath226 , which is ( _ 0,6-_3)m_0,6 + ( _ 0,3-_3 ) m_0,3=(_3 ^ 2 - 1)m_2 .
[ stg:5 ] this can be also obtained by removing the second and third terms in ( [ stg:4 ] ) , as required by the physical relaxation process .
equation ( [ stg:5 ] ) is a standard adiabatic expansion equation with a total energy of @xmath227 . at this stage ,
the only shock that accelerates new electrons is the shock ( a ) , and the emission from it is well described by the standard afterglow theory .
although shocks ( b ) and ( c ) no longer exist , their already accelerated electrons are still present and emitting .
the random lorentz factors in both regions ( 4 and 5 ) , which we still denote as @xmath154 and @xmath219 , evolve as @xmath228 ( fig.4a ) . with the shell - merging dynamics quantified above
, we can calculate the injection signatures in the afterglow lightcurves .
we have developed a code to calculate the dynamics of the shell merging process as well as the emission from various shocks .
since the normal afterglow emission is generally well described by synchrotron radiation , we have neglected the high energy spectral components , such as the electron inverse compton and the emission from the protons . although these components could be important in the mev to gev range ( see zhang & 2001b for a detailed discussion ) ,
usually synchrotron emission is adequate to describe the low energy afterglow ( x - ray , optical and radio ) , at least in the early afterglow phase when injection may be expected . nonetheless , the electron cooling due to the synchrotron self - compton is incorporated through the y parameter ( see , e.g. panaitescu & kumar 2000 ; sari & esin 2001 ;
zhang & 2001b ) .
the random thermal energy in a certain region is an essential parameter in calculating the synchrotron spectrum and the lightcurves . in stages 1 and 2 , the only emitting site is the region heated by the shock ( a ) .
starting from the stage 3 , the emission sites increase to a total number of three , corresponding to the three shocks ( although the shocks ( b ) and ( c ) die out in the stages 5 and 4 , respectively ) .
when the shocks are on , the random lorentz factors at the shocks ( a ) and ( c ) are essentially the lorentz factor of the shocks , i.e. , @xmath229 , and @xmath230 . for the shock ( b ) , since the unshocked region ( region 3 ) is also hot , the random lorentz factor is not simply @xmath173 , but is rather @xmath231 . when the shocks ( b ) and ( c ) die out , the random lorentz factors in the relevant regions are still defined as @xmath232 and @xmath233 , which evolve adiabatically as discussed in 3.3.4 and 3.3.5 . to calculate the synchrotron emission , we assume uniform values of the equipartition factors , i.e. , @xmath234 and @xmath43 , as well as the same electron power law index @xmath235 in the three shock - heated regions .
this is a crude approximation , but it is the simplest case which allows a straightforward comparison of the relative importance of the three shocks . the magnetic field in the region @xmath48 is then @xmath236 . assuming that the injection factor of the electrons is unity , the characteristic injected electron lorentz factor is then @xmath237 ( m_p / m_e ) \gamma_{_{\rm x}}$ ] , where @xmath238 is the random lorentz factor at the region x ( a , b or c ) , @xmath239 is the proton - electron mass ratio .
cooling frequencies are determined from the cooling electron energy , which is defined by the equality between the comoving time and the electron radiative cooling time . for the regions where the shocks are on and keep accelerating electrons ,
the synchrotron emission spectrum is calculated according to the four - segment broken power law description ( sari , piran & narayan 1998 ) . for those regions where
no new electrons are accelerated ( region 5 in the stages 4 and 5 , and region 4 in the stage 5 ) , the synchrotron spectrum is modified by a sharp cut - off above the cooling frequency ( although strictly speaking the cut - off should be exponential ) .
one crucial point in calculating the lightcurves is that the radiation from the three shocks which is emitted at the same fixed - frame time will reach the observer at different observer times .
the main reason is that the region ( 4 + 5 ) usually moves faster ( with @xmath57 ) than the region ( 2 + 3 ) ( with @xmath55 ) , so that for a same time interval in the fixed frame , @xmath240 , the emission from the sites heated by the shocks ( b ) and ( c ) reaches the observer ( apart from the cosmological time dilation effect ) in an interval of @xmath241 , while the emission from the shock ( a ) reaches the observer in an interval of @xmath242 . in our calculations ,
this effect is taken into account approximately . defining the observer s time when the front of shell 3 reaches @xmath116 to be @xmath243
, we can approximate the observer s times corresponding to the regions ( 4 + 5 ) and ( 2 + 3 ) after @xmath243 to be roughly @xmath244 , and @xmath245 , respectively , this effect may be neglected as long as @xmath246 . ] .
we see that the two times are different in the stages 3 and 4 , but are the same in stages 1 , 2 and 5 .
the overall effect of this correction is to `` squeeze '' the lightcurve signal from the fast moving region ( 4 + 5 ) during stages 3 and 4 .
an example of a shell collision is presented in figure 4 . in fig.4a ,
we plot the lorentz factor evolution of various shocks and regions with respect to the blastwave radius , or equivalently the fixed - frame time @xmath247 .
this allows a clear assessment of the five stages analyzed in 3.3 .
figure 4b presents an indicative optical lightcurve , identifying the contributions from the three emitting sites .
the final lightcurve is complicated , with a few features which are caused by different emission sites . to make more sense of this numerical lightcurve
, it is helpful to perform some further analysis .
the emission component from the site related to the shock ( a ) is a standard `` low frequency '' lightcurve ( sari et al . 1998 ) , except for a gentle flattenning around @xmath248 s , which corresponds to the relaxation boosting - up of @xmath55 in the stage 4 ( fig.4a ) .
the emission components corresponding to the shocks ( b ) and ( c ) are more complicated .
the lightcurve slopes in various sections depend on the ( observer s ) time dependence of the characteristic frequency , @xmath249 for slow - cooling ( which is usually the case ) and @xmath250 for fast - cooling , and the peak spectral flux , @xmath251 . here
@xmath49 denotes the bulk lorentz factors ( @xmath55 or @xmath57 ) and @xmath238 denotes the random lorentz factors ( @xmath252 , @xmath154 , and @xmath219 ) .
any change of the scaling law ( with respect to @xmath14 ) of any parameter ( e.g. , @xmath238 , @xmath49 , @xmath45 , @xmath253 , @xmath13 ) results in a break in the lightcurves , and the lightcurve slope also depends on the spectral segment in which the observing frequency lies . in the specific example presented in fig.4b , the climbing - up segment in the lightcurve ( c ) and the flat segment in the lightcurve ( b ) correspond to the violent injection stage 3 , but
the timescale is squeezed due to the increase of @xmath57 with respect to @xmath55 .
the later drop - offs in both lightcurves correspond to the relaxation stage .
finally , all three lightcurves enter the post - relaxation stage at a same observer s time
. there could be more breaks within the same stage .
for example , the break on the lightcurve ( b ) in the post - relaxation stage is due to the transition from the ( 1/3 ) spectral segment to @xmath254 $ ] spectral segment .
two remarks can be made here .
first , although in the stage 5 all the three random lorentz factors decay as @xmath197 , the lightcurve slopes on ( b ) and ( c ) in the @xmath254 $ ] spectral segment are steeper than the slope on ( a ) in the same spectral segment .
this arises from the different scaling law in @xmath255 , more particularly in @xmath94 . for adiabatic expansion in the regions ( b ) and ( c ) , @xmath256 . in the region ( a ) ,
the lightcurve slopes for ( b ) and ( c ) then steepen by 0.75 relative to that of ( a ) spectral regime , which has steepened by itself . ] .
also the whole lightcurves ( b ) and ( c ) during the stage 5 also tilt by 0.75 , which explains why the crossing of @xmath258 in the lightcurve ( b ) looks a little unfamiliar .
second , although the random lorentz factor in the site ( b ) is much higher than those in sites ( a ) and ( c ) , the flux level is not , mainly because the density @xmath259 is smaller than @xmath260 and @xmath261 during the violent collision process . from fig.4b , one can see that the injection signatures of a violent collision are rather complicated
. there could be more than one signature which may be related to different emission sites .
this gives rise to a whole set of injection patterns .
for example , the sharp decline of the site ( c ) emission in the relaxation stage , given an appropriate set of parameters , could be an explanation for the rotse prompt optical flash observation in grb 990123 , which is conventionally interpreted as the cooling from the reverse - shock - heated region in an impulsive shell case ( & rees 1997a , 1999 ; sari & piran 1999 , kobayashi 2000 ) .
the modest sticking - out of the optical lightcurve ( b ) around several @xmath262 s ( fig.4a ) seems to have been observed in some grb lightcurves , although the chromatic nature of this rise rules it out as a possible explanation of the achromatic bump observed in grb 000301c , which can be interpreted as a gravitational micro - lensing event ( garnavich et al .
2000 ) .
the main characteristic which can be used to differentiate a post - injection model from other possible mechanisms leading to bumps in the light curve is the broad - band behavior .
since the injection processes produce global dynamics changes , they will leave imprints at all wavebands . in figure 5 we present the broad - band injection lightcurves for ( a ) a violent injection ; ( b ) a mild injection throughout .
we can see that the mild injection case is similar to the poynting - flux - dominated injection case ( fig.1 ) , as has been discussed in 3.3.2 .
the violent injection case shows rather different and complex behaviors .
the rising times in various bands are essentially the same , but the bump shapes are quite different , mainly because different shocks contribute differently in different bands .
in this paper , we have analyzed the observational consequences of a grb fireball which receives an additional energy injection from the central engine during the afterglow phase .
we have discussed the cases where the injection is either a continuous poynting - flux - dominated wind , or a kinetic - energy - dominated matter shell . for the latter ,
we discuss the cases when the collision is either `` mild '' or `` violent '' , distinguished by whether additional shocks persist and additional emission sites arise in the outflow .
we reach the following general conclusions : \1 .
a post - burst injection energy @xmath263 is generally required to cause a noticeable signature in the afterglow lightcurves .
this is especially so for the poynting flux injection case and for the mild kinetic energy injection case . for a violent collision ,
a faster rear shell ( i.e. @xmath84 ) can compensate for a lower rear shell total energy ( fig.3 ) .
\2 . for the kinetic - energy injection case
, we have explicitly studied the three - shell interaction process , described by a set of equations ( eqs.[[21]]-[[56 ] ] ) .
we find that the formation of an additional pair of shocks from the collision of two shells is not as common as usually thought .
depending on the status of the rear shell when the collision occurs ( spreading or not ) , a minimum relative speed between the two colliding shells is required , as shown in figure 3 . in many cases , such as expected in the internal shock model where slow trailing shells catch up with the fast shells when the latter are decelerated , the relative speed between the shells are usually not high enough , and the injections are likely to be mild .
we expect that a violent injection most likely arises from a late injection of a high entropy shell from the central engine .
since the baryon loading at late times is expected to be smaller than that at earlier times , this is not unusual given a long - lived erratic central engine .
if no additional emission sites form during the injection , such as the case of the poynting - flux - dominated flow , the injection signature is usally very gradual ( fig.1 and fig.5b ) . in the case of a violent injection ,
the signatures are complicated and some could be very abrupt . in either case , the final lorentz factor of the blastwave , as well as the emission flux level , are boosted up with respect to the case without post - injection .
. a prominent characteristic of the post - injection process is the conspiratorial variation of the flux in all bands ( fig.1 and fig.5 ) , due to the changies in the dynamics of the emitting region .
this provides a strong criterion for judging whether a bump signature is due to an injection or to other reasons .
for example , the supernova component ( bloom et al .
1999 ; reichart 1999 ; galama et al .
2000 ) and the dust echo ( esin & blandford 2000 ) interpretations do not predict prominent changes in x - ray and radio lightcurves . the gravitational micro - lensing model may be more difficult to differentiate from a post - injection event , since it is essentially an achromatic feature , although it can have some chromatic features ( granot & loeb 2001 ) . the poynting - flow injection case is also achromatic .
the differentiation between the two possibilities can be only based on the shape of the bump .
for example , the micro - lensing model predicts a more abrupt ( shorter duration ) feature than does the poynting - injection .
another feature of micro - lensing is that the afterglow lightcurve will resume the level of no magnification when the lensing event ends , while in the injection case , the global energetics is always boosted up .
prospects of differentiation look better for a violent matter - injection case , in which the bumps in different bands have quite different shapes , although the rising time is essentially the same ( fig.5a ) .
overall , future broadband afterglow data with very good temporal coverages plus a more detailed modelling can in principle differentiate the injection signature from the other possibilities . within the injection itself ,
different injection signatures may provide direct information about the nature of the injection , and possibly , also about the nature of the central engine .
the treatment presented in this paper is generic and applicable also to more complicated injection processes .
for example , for multi - shell - collision cases , one simply needs to regard the emitting gas after each injection as a new `` impulsive '' component , and a newly injected shell or wind can be treated in the same way . since collisions in the internal shock scenario also involve a similar process ( but without the interaction with the ism )
, the treatment presented here may be also usable in studying grb gamma - ray lightcurves in the prompt phase .
the shell merging is a very complicated process , and here we have used some approximations leading to a semi - analytic description of the whole process . for example , we have assumed that the decelerating shell has a uniform density and speed , while a more accurate treatment ought to take into account the blandford - mckee ( 1976 ) self - similar profiles .
this approximation nonetheless provides a first - order treatment to the problem .
another assumption we have made is that the total internal energy in each region is solely due to the shock heating . strictly speaking , adiabatic loss due to expansion
should be taken into account ( e.g. panaitescu et al . 1998 ) . however , as long as shock heating is going on , the adiabatic loss is not important .
only when shock heating ceases ( e.g. , region 5 in the stages 4 and 5 , and region 4 in the stage 5 ) , do we consider the adiabatic loss .
furthermore , in all our calculations , the effects concerning spreading in the photon arrival time due to the curvature of the shells as well as their non - zero thicknesses have been ignored .
these effects are essential in more detailed afterglow modeling ( e.g. granot , piran & sari 1999 ) , which tend to smear out features on the lightcurves . with these effects taken into account
, we expect that the sharp features shown in fig.5a ought to be smeared out .
finally we compare our results with the previous relevant treatments .
rees & ( 1998 ) , and panaitescu et al .
( 1998 ) have discussed the `` refreshed '' injection scenario in which all the mini - shells are assumed to be ejected simultaneously , but with a power - law distribution of @xmath7 .
although such a scenario is of interest and can be modeled relatively easily , it is unclear whether it is a closer representation of reality .
the poynting - flux - dominated injection within the pulsar central engine model has been discussed earlier by dai & lu ( 1998 ) , but their indicative lightcurve was too abrupt .
their recent treatments ( dai & lu 2001 ; wang & dai 2001 ) generally agree with the gradual evolution picture obtained in this paper . for the shell collision injection case ,
the most relevant precedent discussion is the one by kumar & piran ( 2000 ) . although our results generally agree with theirs on , e.g. , the hydrodynamic treatments , our shock forming condition ( fig.3 ) is more demanding than theirs .
in fact , for the specific example they presented , i.e. , @xmath264 , and @xmath265 , the injection is only mild according to our criterion , while they argue that two shocks will form and propagate into the two shells
. this discrepancy may be due to their not having included the hydrodynamics of the three - shell interaction .
they also presented an indicative injection lightcurve ( their fig.5 ) .
however , this lightcurve is more analogous to our mild - injection signature ( fig.5b ) . as discussed in 3.4 ,
a violent injection tends to generate a more complicated lightcurve with possibly more than one signature , as indicated in fig.4b and fig.5a .
blackman , e. g. , & yi , i. 1998 , apj , 498 , l31 blandford , r , & mckee , c. 1976 , phys .
fluids , 19 , 1130 bloom , j. s. et al . 1999 ,
nature , 401 , 453 chiang , j. , & dermer , c. d. 1999 , apj , 512 , 699 dai , z. g. , & lu , t. 1998 , phys .
, 81 , 4301 . 2001 , a&a , 367 , 501 esin , a. a. , & blandford , r. 2000 , apj , 534 , l151 galama , t. et al .
2000 , apj , 536 , 185 garnavich , p , loeb , a , & stanek , k. 2000 , apj , 544 , l11 granot , j. , & loeb , a. 2001 , apj , 551 , l63 granot , j. , piran , t. , & sari , r. 1999 , apj , 513 , 679 huang , y. f. , dai , z. g. , & lu , t. 1999 , mnras , 309 , 513 kennel , c. f. , & coroniti , f. v. 1984 , apj , 283 , 694 kobayashi , s. 2000 , apj , 545 , 807 kobayashi , s. , piran , t. & sari , r. 1999 , apj , 513 , 669 kumar , p. , & piran , t. 2000 , apj , 532 , 286 lyutikov , m. , & blackman , e. g. 2001 , mnras , 321 , 177 , p. , laguna , p. & rees , m.j .
1993 , apj , 415 , 181 , p. , & rees , m. j. 1997a , apj , 476 , 232 . 1997b , apj , 482 , l29 . 1999 , mnras , 306 , l39 panaitescu , a. , & kumar , p. 2000 , apj , 543 , 66 panaitescu , a. , , p. , & rees , m. j. 1998 , apj , 503 , 314 piran , t. 1999 , phys .
, 314 , 575 piran , t. , shemi , a. & narayan , r. 1993 , mnras , 263 , 861 rees , m. j. & , p. 1998 , apj , 496 , l1 reichart , d.e .
1999 , apj , 521 , l111 sari , r. , & esin , a. a. 2001 , apj , 548 , 787 sari , r. , & piran , t. 1995 , apj , 455 , l143 . 1999 , apj , 517 , l109 sari , r. , piran , t. , & narayan , r. 1998 , apj , 497 , l17 spruit , h. c. , daigne , f. , & drenkhahn , g. 2001 , a&a , 369 , 694 usov , v. v. 1994 , mnras , 267 , 1035 . 1999 , in asp conf .
ser . 190 .
gamma - ray bursts : the first three minutes , ed .
j. poutanen & r. svensson ( san francisco : asp ) , 153 wang , w. & dai , z.g .
2001 , chin .
, 18 , 1153 zhang , b. , & , p. 2001a
, apj , 552 , l35 . 2001b , apj , 559 , 110 | we consider generically the problem of a gamma - ray burst ( grb ) fireball with an additional energy injection , either in the form of a poynting - flux - dominated outflow or a kinetic - energy - dominated matter shell injected after the burst .
generally , a total injection energy comparable to that of the impulsive energy in the initial fireball is required to make a detectable signature in the afterglow lightcurves . when this criterion is met in the case of poynting - flux - dominated injection
, this leads to a gradual achromatic bump appearing in the otherwise power - law afterglow lightcurve .
alternatively , in the case when the injection is kinetic - energy - dominated , the results depend on whether the collision between the rear ( injected ) and the forward shell is mild or violent .
if the relative velocity between the colliding shells does not exceed a critical value defined by their energy ratio , the collision is mild , and the injection may be analogous to the poynting - flux injection case . otherwise , the injection is violent , and an additional pair of shocks will form at the discontinuity between two colliding shells , so that there are altogether three shock - heated regions from which the emission contributes to the final lightcurves .
we describe the shell - merging process in detail including collision and relaxation by taking into account the dynamical evolution and the emission from the various shocks involved .
assuming synchrotron emission , we calculate afterglow lightcurves in the x - ray , optical and radio bands for the various cases .
the injection signatures due to violent matter - dominated collisions are abrupt and complicated , due to the emission from any of the three emitting regions , and depending on the injection parameters and the observed energy bands .
this differs from the gradual bump signature found in the poynting - flux injection case . in both the poynting - flux - dominated and the kinetic - energy - dominated cases ,
the energetics of the fireball as well as the absolute afterglow flux level after the injection are boosted with respect to the one without post - burst injection . identifying the different injection signatures from future early afterglow observations
may provide diagnostics about the nature of the fireball and of the central engine . |
despite the terrifying fact that numerous operational fission or even thermonuclear bombs exist on our planet , there is a great interest in the basic principles and physics underlying the concept of nuclear weapons . during the last decades
, an increasing number of details concerning the design of the gadget , the first atomic bomb ignited in the trinity test , have been revealed . in the present work ,
the first man - made nuclear explosion is modeled in a simplified but instructive manner , which nevertheless leads to realistic and even rather accurate quantitative results .
the calculations are based on a point model in the analogous sense of nuclear reactor point models or neutron point kinetics , where the whole structure of the reactor core is averaged out in an effective way , removing the spatial structure of the object under study which therefore becomes a structureless point , but still retaining the basic physical features in the time domain .
below , we will model a exploding plutonium core including its surrounding matter both as a sphere or a ball with a time - dependent radius @xmath0 , where the expansion is driven by ( a gradient of ) the radiation pressure produced by the energy released by the nuclear chain reaction ( see also fig .
( [ fig_model ] ) ) . on the one hand , the time - dependent density of fissile @xmath1pu atoms @xmath2 and other crucial quantities are assumed to be spatially constant inside the sphere as an approximation and one defining feature of the present point model . on the other hand
, we will assume that the explosion builds up a fireball , a matter- , or a blast- shell confined by a wall of fire in the close vicinity of the surface of the sphere of radius @xmath0 , containing all the material that originally was located inside the sphere finally compressed to a thin layer .
the interior of the fireball is basically matter - free but filled with black body radiation with a temperature @xmath3 , i.e. with a homogeneous photon gas reaching a very high energy density and a radiation pressure of several hundred gigabars during the nuclear explosion .
this picture is certainly justified within the first microseconds after ignition of the chain reaction by a neutron source , where the explosion generates very high temperatures of several @xmath4 @xcite , but one must keep in mind that applying concepts from equilibrium thermodynamics clearly represents an approximation . since the mean free path of the photons inside the dense plutonium core is of the order of @xmath5 , the major part of the energy released by fission will remain in the core during this phase .
the radiation temperature must be clearly distinguished from the lower temperature distribution inside the expanding matter shell , whose detailed structure is of minor interest in the present exposition .
however , the hot matter shell emits hazardous intense x - rays , ultraviolet light , visible light , infrared light , and thermal radiation to the environment . due to its symmetries , the point kinetic model shares some similarities with cosmological models of the early universe . , including all the matter originally located inside the sphere with radius @xmath0 .
( b ) the expansion of the core is driven by the radiation pressure inside the matter - free sphere , which converts the energy released from fission to the matter s kinetic energy .
( c ) the dynamics of the chain reaction is inspired from ordinary neutron kinetics and transport theory inside a homogeneous plutonium sphere.,width=234 ] the time - dependent diffusion equation for the prompt neutron flux density @xmath6 or the neutron density @xmath7 inside the sphere containing the fissionable material ( @xmath8 reads @xmath9 where the neutron density @xmath10 is related to the neutron flux density @xmath11 by the average fission neutron velocity .
the fast neutron spectrum is not known with high accuracy , and the neutrons inside the sphere containing the fissionable material undergo elastic but also inelastic scattering .
below , we will work with a generally accepted value for average neutron velocity of @xmath12 . a diffusion model according to eq . ( [ diffusion_model ] )
is justified to some extent since the velocity of fission neutrons is generally much bigger than the velocity of the fissionable material . according to diffusion theory , a gradient of the neutron flux density induces a neutron current density @xmath13 according to fick s law of diffusion @xmath14 where @xmath15 is the diffusion constant for fast neutrons inside the material under study ,
i.e. @xmath16 is a parameter which basically describes the local interactions of the neutrons governing how easily the neutrons can move .
the divergence of the current density @xmath13 equals the local neutron leakage rate per volume due to diffusion @xmath17 \ , .\ ] ] as already mentioned we presume homogeneity in the sense that the diffusion constant @xmath18 and the total , fission , absorption , capture , or scattering macroscopic cross sections @xmath19 are assumed to be spatially constant inside the sphere where the diffusion equation will be investigated .
a macroscopic cross section @xmath20 is the collective cross section of all atoms per volume .
this collective cross section density gets permeated by the neutron flux density @xmath6 ; accordingly , the corresponding reaction rate per volume is given by the product @xmath21 .
given the average number @xmath22 of ( prompt ) neutrons released per fission , the local neutron production rate per volume is @xmath23 and since the material inside the core absorbs neutron at a rate per volume @xmath24 one finally arrives at the spatially homogenized version of the balance equation ( [ diffusion_model ] ) @xmath25 @xmath26 of course , quantities like the density of atoms @xmath27 and macroscopic cross sections directly depend on the time - dependent compression factor @xmath28 or the corresponding density of the sphere , where @xmath29 is the radius of the uncompressed plutonium core .
the ansatz @xmath30 with the inverse bomb period @xmath31 ( rossi alpha ) @xmath32 leads to @xmath33 in the spherically symmetric case , the laplacian @xmath34 becomes @xmath35 and making the * q*uasi-*s*tatic ansatz where one neglects the last term in eq .
( [ separation ] ) one finds the well - known solution which is an eigenstate of the laplacian @xmath36 where @xmath37 is the so - called geometric buckling to be specified below , and from eq .
( [ separation ] ) and @xmath38 follows the inverse bomb period @xmath39 and the so - called material buckling is given in accordance with the literature by @xmath40 with @xmath41 .
space and time arguments have been omitted for the sake of notational simplicity . above , the infinite multiplication factor for a core without leakage @xmath42 has also been introduced .
now two comments are in order .
first , the separation ansatz in eq .
( [ sep_ansatz ] ) is certainly not unique . a separation la @xmath43 , such that obviously @xmath44 , is impossible since the diffusion equation under study is considered on a time - dependent domain @xmath45 , and one should note that @xmath18 and @xmath46 are time - dependent quantities in @xmath47 also whose evolution in time will be governed by the dynamics induced by the chain reaction to be discussed further below .
second , we need a strategy to calculate the geometric buckling appearing in the approximate ansatz ( [ nf_ansatz ] ) for @xmath48 . from diffusion theory
one learns for the time - independent case that the spherically symmetric neutron ( flux ) density inside a homogeneous ball with radius @xmath49 surrounded by a vacuum is maximal in the center ( @xmath50 ) and minimal at the boundary ( @xmath51 ) where the neutrons leak out .
if @xmath52 for @xmath53 were true , the geometric buckling would be given by @xmath54 , so that @xmath55 vanishes .
however , according to a common approximation inspired from diffusion theory , the neutron ( flux ) density vanishes on an imaginary sphere outside the core with a so - called extrapolated radius @xmath56 to be calculated below . with this boundary condition applied to the time - dependent case ,
the geometric buckling becomes @xmath57 and @xmath31 can also be computed .
integrating over the ball @xmath47 , we obtain from eq .
( [ separation ] ) , disregarding the last term on the right , for the neutron number @xmath58 and consequently @xmath59 the fact that the last term in eq .
( [ separation ] ) @xmath60 is negligible after integration represents an adiabatic approximation which can be justified by the observation that the neutron dynamics is governed by a much smaller time scale than the expansion of the plutonium core .
the expansion destroys the separability of the diffusion equation . after having defined how to calculate the quasi - static neutron ( flux ) distribution and the geometric buckling in the expanding core as a quasi - static object , eq .
( [ diffeq_period ] ) can be inferred as an approximation to the true neutron number dynamics . in the presence of a neutron source ( due to spontaneous fission of @xmath61pu or an alpha - beryllium source ) ,
( [ growth ] ) can be equipped with a source term @xmath62 @xmath63 using now the extrapolated radius @xmath64 inspired from neutron transport theory @xcite where the extrapolated neutron flux density eq .
( [ nf_ansatz ] ) is assumed to vanish @xmath65 leads in the truly static case ( @xmath66 ) to the condition for the critical extrapolated radius @xmath67 which is related to the geometric buckling introduced above @xmath68 for illustrative purposes , we readily calculate the critical mass @xmath69 for pure plutonium in the alpha phase .
the neutron cross sections averaged over the fission spectrum are given by @xmath70 , @xmath71 , @xmath72 , @xmath73 , and @xmath74 .
the average number of fast neutrons released in a fission induced by fast fission neutrons is @xmath75 , delayed neutrons included in the stationary case . in the alpha phase , the density of plutonium is @xmath76 , from the isotope mass @xmath77 one calculates the density of plutonium atoms @xmath78 with the avogadro constant @xmath79 . from diffusion theory follow the approximate expressions for the diffusion constant @xmath16 and the transport mean free path @xmath80 @xmath81 with an approximate value of the average cosine of the neutron scattering angle @xmath82 , where @xmath83 is the atomic mass number .
one finally obtains @xmath84 corresponding to a critical mass of @xmath85 an analogous calculation for pure @xmath86u with @xmath87 , @xmath88 , @xmath89 , @xmath90 , @xmath91 , @xmath92 , @xmath93 , and @xmath94 gives @xmath95 and @xmath96 .
analogous calculations concerning the critical mass of uranium can also be found in @xcite .
interestingly , doubling the density of @xmath97pu by compressing it by a compression factor @xmath98 reduces the critical radius to @xmath99 and the critical mass to @xmath100 .
adding an extended neutron reflecting tamper around the plutonium core also lowers the critical mass . in this case ,
the spherically symmetric neutron flux density in the tamper behaves like @xmath101 , and the critical mass can be calculated from the critical condition that the stationary diffusion current density is continuous everywhere .
we want to generalize the preliminary considerations above to a homogeneous @xmath97pu sphere with a time - dependent radius @xmath0 containing @xmath1pu with a total mass @xmath102 .
it is straightforward to calculate relevant quantities like macroscopic cross sections discussed above from convenient quantities like the number of @xmath97pu atoms @xmath103 , the volume @xmath104 of the sphere or the corresponding atomic density @xmath2 @xmath105 where @xmath104 is the momentary volume of the sphere .
before the chain reaction starts , the mass and the number of @xmath97pu nuclei is given by initial values @xmath106 and @xmath107 , respectively . from @xmath0 and @xmath108
follows the extrapolated radius @xmath109 and the corresponding material and geometric bucklings @xmath110 ; in the case of the geometric buckling , the quasi - static approximation is @xmath111 such that @xmath112 .
the inverse bomb period @xmath113 follows from eq .
( [ ibp ] ) and serves for the update of the neutron flux density inside the sphere . with respect to the inverse bomb period ,
it is useful to introduce the ( time - dependent ) prompt neutron generation time @xmath114 the effective neutron multiplication factor @xmath115 and the reactivity @xmath116 @xmath117 which are related to the inverse bomb period via @xmath118 where @xmath119 is the average neutron lifetime in the plutonium assembly .
then the @xmath120 neutrons inside the sphere generate an average flux density of @xmath121 and the power released inside the sphere is @xmath122 where @xmath123 is the average energy released per fast fission ( anti - neutrinos neglected ) .
the expansion of the sphere is driven by radiation pressure in the hot phase of the explosion .
assuming that the released energy @xmath124 is converted into black body radiation and kinetic energy of the expanding matter shell , one has an average pressure inside the sphere @xmath125 with the stefan - boltzmann constant @xmath126 and the speed of light @xmath127 .
@xmath128 is the photon energy stored inside the sphere .
also the decrease of the number @xmath129 of @xmath97pu atoms should be taken into account @xmath130 since the pressure gradient drives the expansion of the sphere , one has to make a reasonable assumption at the present stage of the model concerning the behaviour of the pressure gradient . however , since the pressure will create a fireball filled with a photon gas pushing away all the matter from the explosion center , the corresponding radiation energy is converted into kinetic energy of the matter , and an ansatz describing the work done by the radiation pressure on the surrounding and moving matter distributed over an area of @xmath131 according to @xmath132 is reasonable , where @xmath133 is the mass of all the matter that was located inside a sphere with radius @xmath0 before the nuclear detonation and that got compressed onto a spherical shell with approximately the same radius .
accordingly , the photon energy inside the sphere is given by @xmath134
since in the trinity test , the plutonium core was ( probably ) surrounded by a @xmath135 @xmath136u - tamper , a @xmath137 aluminum pusher shell plus @xmath138 additional material including the high explosives surrounding the shells around the core , @xmath139 was modeled by @xmath140 @xmath141 @xmath142 @xmath143 with the minimum radius @xmath144 of the imploded core , a tamper radius @xmath145 , an aluminum pusher shell radius @xmath146 , an approximate bomb radius when the nuclear detonation starts @xmath147 , and the density of air @xmath148 .
above , approximate ( de)compression effects of the bomb material when reaching a maximum core compression factor @xmath149 have been taken into account . the small neutron source located in the center of the plutonium core starting the chain reaction at maximum compression
was assumed to emit enough neutrons of the order @xmath150 such that the chain reaction is initiated immediately when the shock wave of the conventional explosion reaches the center of the core .
since delayed neutrons play no role on the time - scale of a nuclear explosion , an average number of @xmath151 prompt neutrons per fission was used for the simulation .
using super - grade plutonium , the probability that the chain reaction starts due to spontaneous fission of @xmath61pu is rather small for core compression times of the order of @xmath152 .
the trinity plutonium core consisted of @xmath153 plutonium - gallium alloy containing @xmath154 gallium with a fast neutron scattering cross section of @xmath155 .
the plutonium phase with the lowest density is the delta phase , with theoretical density of @xmath156 .
although existing only in the temperature range of @xmath157 , it can be stabilized at room temperature by adding small quantities of gallium to the plutonium .
the convenient non - metric units of a kiloton ( @xmath158 fission events ) and a shake ( @xmath159 ) will also be used in the following @xcite .
most of the results presented in this section were calculated for a chain reaction starting at the time of maximum core compression with a compression factor of @xmath160 @xcite .
adaptive timesteps of the order of @xmath161 have been used to model the actual explosion based on a simple euler discretization of the dynamical equations presented in the last section .
one should note that it does not make too much sense to calculate an extrapolated radius for the neutron flux density outside the plutonium core for the uranium tamper during the intense phase of the nuclear explosion , since the average lifetime @xmath162 of the neutrons in the @xmath163u tamper is much bigger than in the core .
the lifetime @xmath162 is given by the average length of the path @xmath164 a neutron travels in the uranium until it gets absorbed there by fission or capture @xmath165 divided by the neutron velocity .
the corresponding cross sections averaged over the fission spectrum are @xmath166 and @xmath167 , and from the uranium density @xmath168 at room temperature and pressure and a compression factor @xmath149 one calculates @xmath169 .
therefore , the neutrons need a time to adapt their distribution which is of the order of the time the main part of the energy is released by the nuclear explosion .
in fact , it turns out that simply using the extrapolated length for a core surrounded by a vacuum calculated from eq .
( [ extrapol ] ) already leads to reasonable explosion yields for trinity , as depicted in fig .
( [ fig_compression_yield ] ) . for a compression factor of @xmath170 ,
the yield is @xmath171 kilotons .
however , in order to have consistency with some data used in the literature , the constant @xmath172 in eq .
( [ extrapol ] ) was slightly reduced to @xmath173 in order to gauge our model to a yield of @xmath174 for @xmath175 .
since the limits of the physical validity of diffusion theory are reached and since the point kinetic model is based on some strong homogeneity assumptions concerning the neutron distribution inside the core in the prompt supercritical phase and the idea of a spherical shell - like structure of the fireball , such a strategy may be acceptable .
a yield of 15kt is the currently accepted value for the trinity test due to the @xmath97pu content in the gadget only ; still , it is estimated that the @xmath176u tamper additionally released about @xmath177 , but we do not try to model this surplus in this study which finally leads to the generally accepted value for the total trinity yield of of @xmath178 .
of course , the less relevant energy production by the tamper could also be assessed in our approach by modeling the prompt neutron leakage from the core into the tamper . in the _
static _ and homogeneous case , the neutron leakage rate @xmath179 is given by @xmath180 with the geometric buckling @xmath181 fig .
( [ fig_trinity_expansion ] ) displays the size of the fireball as a function of the time after ignition .
the results are in very good agreement with the actual radii measured in the trinity test - see , e.g. , fig .
( [ fig_trinity ] ) showing a high - speed rapatronic camera photograph of the fireball taken @xmath182 after ignition @xcite .
one should note at this stage that the total yield @xmath183 of the nuclear explosion can be decomposed in a more elaborate model into different parts .
e.g. , within a millionth of a second after the explosion , all matter including the bomb itself and the surrounding air is transformed into a very hot plasma which emits thermal radiation as x - rays , which again gets reabsorbed by the dense shock front of the fireball itself .
after some seconds , the energy of the explosion can be decomposed into the blast energy @xmath184 , i.e. the kinetic energy transferred primarily to the air ( still about @xmath185 ) , thermal radiation @xmath186 including light ( @xmath187 ) , and nuclear radiation @xmath188 of various types ( @xmath189 ) . at this time
, the focus of investigations in the literature is rather on the damage done to humans and the environment and the present considerations must be replaced by different physical and ethical concepts @xcite , @xcite .
( [ fig_upshot_grable ] ) shows a snapshot taken in 1953 shortly after ignition of the device used in the nuclear weapons test upshot - knothole grable , where an estimated 15kt gun - type fission bomb exploded @xmath190 above ground , producing finally a spherically symmetric fireball , despite the asymmetric aspects of the pre - ignition device @xcite . supposing that a considerable part @xmath184 of the total yield @xmath183 of an atomic bomb is converted into the kinetic energy of the matter moving at a distance @xmath0 from the explosion center with a velocity @xmath191 , and neglecting the mass of the bomb for fireballs with a radius larger than @xmath192 and the energy of the radiation inside the ball , one has from @xmath193 the kinetic energy which equals the blast energy @xmath194 and therefore @xmath195 integrating this expression leads to @xmath196 with some integration constant @xmath197 , and finally one has @xmath198 for large radii , the integration constant is negligible and eq .
( [ blast_radius ] ) is in good agreement with observations from nuclear tests and in very good agreement with the numerical results presented in fig .
( [ fig_trinity_expansion ] ) .
one also observes that doubling the blast yield by a factor of two increases the radius @xmath199 of the fireball at the same time after ignition only by @xmath200 .
it is also instructive to consider the fireball expansion velocity as a function of the distance from the explosion center .
the tail for larger distances in fig .
( [ fig_exp_velocity ] ) corresponds to the description via eq .
( [ blast_radius ] ) . after a first acceleration phase due to the violent energy release by the plutonium core
, the expansion enters a deceleration phase caused by the high density of the uranium tamper .
when the explosion enters the aluminum with a lower density , the pressure is still high enough to initiate a reacceleration of the expansion .
finally , the rest of the bomb and the air surrounding the device are pushed away , still at a speed of the order of @xmath201 .
( [ fig_trinity_temperature ] ) displays the radiation temperature reached inside the trinity fireball .
temperatures of the order of @xmath202 correspond to particle energies of @xmath203 with boltzmann s constant @xmath204 .
the temperature for complete ionization of plutonium is of the order of some @xmath205 and the binding energy of the innermost k - shell electrons is larger than @xmath206 , therefore , in the plasma of a fission explosion one is still far away from complete ionization , but the plutonium atom may lose a large fraction of its electrons for a very short period of some few shakes .
note that for partially ionized plasma the degree of ionization can be calculated from the saha equation @xcite .
an analysis of the numerical results of the point kinetic model confirmed that the total yield @xmath183 of the explosion of a plutonium implosion device is approximately proportional to the inserted reactivity @xmath207 , i.e. , the reactivity of the compressed core when the chain reaction starts , as anticipated by serber @xcite .
roughly , one has in the present model @xmath208 of course , the total yield depends on the physical parameters governing the point kinetic model .
the dependence of the yield is not very pronounced for variations of @xmath209 .
if more energy would be released per single fission , the core would expand faster and reach second prompt criticality within a shorter period , which has a negative impact on the total yield of the device .
however , it is advantageous to have fast neutrons , large fission cross sections , and a large fast fission neutron yield @xmath22 .
the fraction of plutonium nuclei undergoing fission during a nuclear explosion is called the efficiency of the bomb .
only about @xmath210 of the plutonium in the core was destroyed in the trinity test . in the vast majority of cases ,
a plutonium nucleus splits into two smaller nuclei and 2,3 , or 4 additional neutrons . only about @xmath211 of fissions
are ternary fissions , producing a third light nucleus such as @xmath212he ( @xmath213 ) or @xmath214h ( @xmath215 ) .
e.g. , fissioning 1000 @xmath97pu nuclei produces on an average @xmath216 fission fragments with an atomic mass number of @xmath217 - namely @xmath218sb , @xmath219te , and @xmath220i .
these fragments are neutron - rich and tend to undergo subsequent @xmath221-decays according to the decay chain with given half - lives @xmath222 until a ( quasi ) stable fission product like @xmath223cs with a half - life of @xmath224 years is reached . during a nuclear explosion ,
such @xmath221-decays of fission fragments play no role due to the very short time - scale of the nuclear chain reaction .
still , the initial fission fragments have some influence on the chain reaction . because they are neutron - rich , their absorptive properties are rather irrelevant , but the fragments act as additional scatterers to the neutrons , influencing thereby the neutron transport inside the core .
scattering has a confining effect , since the neutron diffusion constant decreases when the neutron scattering cross section increases .
the scattering cross sections of the fission fragments were not taken into account in the present simulation , but including a corresponding macroscopic scattering cross section term in the point kinetic model is straightforward .
note , however , that xenon-135 has the highest known thermal neutron absorption cross section of any nuclide , namely @xmath225 for neutrons with a kinetic energy of @xmath226 . in nuclear reactors , @xmath227xe can strongly influence the reactivity balance , but its concentration typically varies on a time - scale of the order of some hours .
reactor grade plutonium contains different plutonium isotopes . whereas the heat generated by the @xmath176pu is a problem for the integrity of the explosives surrounding the nuclear part of the bomb , @xmath61pu is a source of fast neutrons which can start the chain reaction before the plutonium core has reached sufficiently high compression for the intended efficiency .
@xmath61pu undergoes @xmath228 spontaneous fissions per gramsecond , releasing neutrons at a rate of about @xmath229 @xcite . in most cases ,
the splitting nucleus emits one , two , or three neutrons with comparable probabilities .
when a neutron is released in the core , there is still some probability that it escapes the fissile zone .
the probability that a fast neutron triggers a fission is given by @xmath230 .
in the limit of an infinitely extended reactor , this probability becomes @xmath231 . in a critical @xmath97pu assembly
where @xmath232 , one has @xmath233
, i.e. only one third of the neutrons initiates a fission which releases about three new neutrons on an average .
a small part of the neutrons will be captured , producing thereby @xmath61pu , but most of the neutrons diffuse out of the core .
this fact has to be taken into account when modeling fizzle probabilities . for the sake of clarity
, we mention here that fissile materials can sustain a chain reaction with neutrons of any energy , whereas fissionable materials are materials that can only be made to fission with fast neutrons .
fertile materials are materials that can be transformed ( i.e. , transmuted ) into fissile materials by the bombardment of neutrons inside a reactor . in this sense , @xmath61pu is fissionable and fertile .
( [ fig_trinity_fizzle ] ) shows the yield for 700 trinity explosions with randomly chosen neutron source terms corresponding to a @xmath61pu content ranging from @xmath234 ( super grade to mox grade ) in the trinity core .
the maximum compression factor reached has been randomly blurred in order to mimic different efficiencies of the high explosives and to render the plot more legible to the eye . for fig .
( [ fig_trinity_fizzle ] ) , a compression time of @xmath235 was assumed from first prompt criticality until maximum compression without predetonation @xcite , corresponding to a velocity of the order of @xmath236 of the surface of the imploding plutonium core .
the situation changes when the insertion time is doubled , since the fizzle probability is higher then , as depicted in fig .
( [ fig_trinity_fizzle_slow ] ) .
note that for the simulation of the trinity device , it was assumed that the chain reaction starts at full compression when the core is basically at rest .
for fizzles , the contraction phase has to be modeled during which the core reaches first prompt criticality where @xmath66 .
shortly after having reached the maximum reactivity @xmath207 , the expansion phase starts and the core eventually reaches second prompt criticality where @xmath66 again .
pu content ( trinity type device).,width=415 ] one should note that even when the trinity device fizzles in the most unfavorable manner ( from the war strategic view ) , it still produces an energy of the order of @xmath237 or more , enough to destroy and contaminate a city district .
when the chain reaction starts at a very low positive prompt reactivity , the core still has same time to further contract , eventually leading to a considerable energy release .
the simulations show that the fizzle yield scales as @xmath238 , where @xmath239 is a typical initial compression velocity of the core when reaching prompt criticality . however , precise estimations concerning the minimal fizzle yield depend strongly on how the implosion of the core is modeled . in fig .
( [ fig_trinity_fizzle_slow ] ) , in order to facilitate the probabilistic comparison with fig .
( [ fig_trinity_fizzle ] ) , it was assumed that the homogeneously contracting core has a higher compressibility and still reaches the full compression factor @xmath175 when the bomb does not fizzle , although the initial contraction velocity is lower .
in reality , the inserted reactivity would decrease for a lower compression velocity , leading to a corresponding reduction of the maximum yield of the device . the present work is not inteneded to model implosion scenarios , which belong to the harder part from the physical point of view .
in addition , the pressure in the hot phase of the fizzle explosion is no longer radiation but matter dominated for low yields , and the limits of our model are reached .
of course , the transition to the low temperature domain where the gas pressure @xmath240 becomes dominant could also be integrated in the point kinetic model . to take it with a grain of salt
, one may argue that it is no longer relevant for the construction of a fizzle bomb with low quality plutonium to integrate an efficient neutron source in the core , but to achieve a fast compression of the core with efficient high explosives .
pu content ( trinity type device ) for slow reactivity insertion.,width=415 ] furthermore , a high @xmath61pu content is disadvantageous for the ( fizzle ) yield since the fission cross section of @xmath61pu averaged over the fission spectrum is @xmath241 @xcite , whereas for @xmath97pu one has @xmath242 .
the impact of @xmath243pu is relatively small .
the construction of a bomb with low quality plutonium would therefore involve high costs and comparatively low explosive yield .
it is a striking fact that sophisticated simulations of nuclear explosions do not necessarily lead to better results from a numerical point of view for many characteristic quantities than an analysis based on high school mathematics like the one presented in this work .
this is due to the complexity of the non - equilibrium thermodynamics involved in nuclear explosions and the lack of some specific exact experimental data .
the present paper is a reminiscence of robert serber s lectures @xcite which were given in 1943 to new members of the manhattan project with the aim to explain the basic scientific facts of the wartime enterprise , and which assembled in note form and mimeographed became the legendary la-1 , the los alamos primer , classified secret - limited for twenty years after the second world war . despite its simplifications ,
the point kinetic model provides a flexible approach to the physics of the uncontrolled chain reaction .
performing the corresponding simulations is straightforward and may help students to interpret the extreme physical conditions that briefly occur within nuclear fission bombs .
the influence of engineering details like tampers and other material surrounding the core can be studied .
effects which are relevant , e.g. , for discussions concerning the non - proliferation of nuclear weapons , can be accessed with ease .
one - dimensional numerical simulations beyond the point kinetic model including radiative transport phenomena for the spherically symmetric case will be presented in a forthcoming paper .
the lesson that relatively simple toy model calculations are sufficient to get approximate estimates of the yields of nuclear detonations does not constitute a real security problem . the main obstacle for terrorist organizations to build an atomic bomb is the acquisition and technical handling of the potentially highly radiotoxic material with an acceptable isotopic vector needed for an efficient device .
however , for developing countries , such problems will play an increasingly smaller role in the near future .
possession of nuclear weapons may prevent states from entering into destructive wars .
emerging decadent societies in decadent democracies , which are already in possession of numerous nuclear weapons and which are no longer able to elect wise people as their leaders or to control their nuclear inventory , may initiate the final countdown to zero for the third atomic bomb that will be deployed in an act of war .
in such a case the question remains whether the people are still able to apply ideas from the age of enlightenment , like they are found in the declaration of independence , urging the people to act .
for the sake of future generations , it is the duty of every person never to forget about and to warn of the terrible destructive force of ( thermo-)nuclear and other weapons of mass destruction .
the author wishes to thank josef ochsner from the paul scherrer institute in switzerland for valuable comments and carefully reading the manuscript .
m. gttsche , g. kirchner , _ improving neutron multiplicity counting for the spatial dependence of multiplication : results for spherical plutonium samples _
instr . meth .
res . a * 798 * ( 2015 ) 99 - 106 . | a concise point kinetic model of the explosion of a prompt supercritical sphere driven by a nuclear fission chain reaction is presented .
the findings are in good agreement with the data available for trinity , the first detonation of a nuclear weapon conducted by the united states army as part of the manhattan project .
results are presented for an implosion device containing pure plutonium-239 , although the model can be easily applied to , e.g. , uranium-235 .
the fizzle probability and corresponding yield of a fission bomb containing plutonium recovered from reactor fuel and therefore containing significant amounts of spontaneously fissioning plutonium-240 which can induce a predetonation of the device is illustrated by adding a corresponding source term in the presented model .
related questions whether a bomb could be made by developing countries or terrorist organizations can be tackled this way .
although the information needed to answer such questions is in the public domain , it is difficult to extract a consistent picture of the subject for members of organizations who are concerned about the proliferation of nuclear explosives .
+ 0.1 cm * physics and astronomy classification scheme ( 2010 ) . * 28.20.-v neutron physics ; 25.85.ec neutron - induced fission ; 28.70.+y nuclear explosions . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.